proyecto/notebook_speech_to_command_IA2.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Oficial | Proyecto_IA2_SPEECH_COMMAND.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU",
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "23c1f6a1983944a8951eb86aaee87a97": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_1a6e07770c564ab1921df9fc852b3e5b",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_0ac2ac9530ec4eb29d30a0565761e691",
              "IPY_MODEL_74b0ce2b527344b6830f54333a54f21e"
            ]
          }
        },
        "1a6e07770c564ab1921df9fc852b3e5b": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "0ac2ac9530ec4eb29d30a0565761e691": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_1c7f757b3ceb463f946ee4e1767713b5",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 650,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 650,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_e32333d33f2346b2a2c7a1e5e2e9f57d"
          }
        },
        "74b0ce2b527344b6830f54333a54f21e": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_5cc377511d7e4142a9321b6016b69942",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 650/650 [1:56:04&lt;00:00, 10.71s/it]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_439299fc8fd84a9e85f67afe80d7d666"
          }
        },
        "1c7f757b3ceb463f946ee4e1767713b5": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "e32333d33f2346b2a2c7a1e5e2e9f57d": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "5cc377511d7e4142a9321b6016b69942": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "439299fc8fd84a9e85f67afe80d7d666": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "965d954a066c4acebdbc404e5362a5c5": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_37e475f586a5434db67ffae8c9646989",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_2d3a9ce70da74f81812eddc03e4834c7",
              "IPY_MODEL_73e1ddb9cb744e5b8754a435d2b41577"
            ]
          }
        },
        "37e475f586a5434db67ffae8c9646989": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "2d3a9ce70da74f81812eddc03e4834c7": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_8bd7fb2857114340a772fdab802005b3",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 650,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 650,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_c94eef4b6a3d4af48ee4467bb578bc42"
          }
        },
        "73e1ddb9cb744e5b8754a435d2b41577": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_5e9d515965d44c019205e5fd782a2326",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 650/650 [1:48:56&lt;00:00, 10.06s/it]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_081a5140391047d5808e9fa83a209eee"
          }
        },
        "8bd7fb2857114340a772fdab802005b3": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "c94eef4b6a3d4af48ee4467bb578bc42": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "5e9d515965d44c019205e5fd782a2326": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "081a5140391047d5808e9fa83a209eee": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "c31d3ce029084266a4240f8bd2d57430": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_f0e93407f8414a51b0114f8e1a69d3f9",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_dbcc3cb7088e48a2a7895da21bf244bd",
              "IPY_MODEL_0cfc5bb596cf44cbaad01f09e647f748"
            ]
          }
        },
        "f0e93407f8414a51b0114f8e1a69d3f9": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "dbcc3cb7088e48a2a7895da21bf244bd": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_823dcaccc7904a61a32e034c9589a689",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 650,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 650,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_afb64f031f2d4af3975309620b643185"
          }
        },
        "0cfc5bb596cf44cbaad01f09e647f748": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_1738aa539ac842ec912041b0e5599c03",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 650/650 [1:41:53&lt;00:00,  9.41s/it]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_d072fa81c5344451a9f422200ea2e5d2"
          }
        },
        "823dcaccc7904a61a32e034c9589a689": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "afb64f031f2d4af3975309620b643185": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "1738aa539ac842ec912041b0e5599c03": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "d072fa81c5344451a9f422200ea2e5d2": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "49d69c43b6ef4dff8ecec9f9469e104d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_d89ca8bcc8db4cb9826a9f3baecc9ad2",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_9c417a0c2f0741cc87cc97f0edb1530f",
              "IPY_MODEL_1c40d4b3feff4cbb922069ccee260ad0"
            ]
          }
        },
        "d89ca8bcc8db4cb9826a9f3baecc9ad2": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "9c417a0c2f0741cc87cc97f0edb1530f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_52b5d0b5c03b42ccb45777db7e726d40",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 650,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 650,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_18b07cac759640f083c3aee38b98e249"
          }
        },
        "1c40d4b3feff4cbb922069ccee260ad0": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_b9edb41cc3bb453ab5e377fc446c0fd1",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 650/650 [1:35:00&lt;00:00,  8.77s/it]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_2b6bf280dbad4598adfb76baa563c3e2"
          }
        },
        "52b5d0b5c03b42ccb45777db7e726d40": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "18b07cac759640f083c3aee38b98e249": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "b9edb41cc3bb453ab5e377fc446c0fd1": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "2b6bf280dbad4598adfb76baa563c3e2": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "adbb91f1450847faa861d44ed598c204": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_f6105d82b52843bba7f9c1c0021067b2",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_bc2f1fbdf34e4435b4fc4b45b54b5d79",
              "IPY_MODEL_00a58be961a2458498e34b4f3e6b5ea6"
            ]
          }
        },
        "f6105d82b52843bba7f9c1c0021067b2": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "bc2f1fbdf34e4435b4fc4b45b54b5d79": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_c987d7bfe36046f6b9b8e8c7bf01bf3e",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 650,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 650,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_b3d0c9cf18394831a7c921780531c6cc"
          }
        },
        "00a58be961a2458498e34b4f3e6b5ea6": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_b05650179ffd4323b4441e3a48120ea6",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 650/650 [1:28:02&lt;00:00,  8.13s/it]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_7f73c44b92714c5599ca7af9f492a226"
          }
        },
        "c987d7bfe36046f6b9b8e8c7bf01bf3e": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "b3d0c9cf18394831a7c921780531c6cc": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "b05650179ffd4323b4441e3a48120ea6": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "7f73c44b92714c5599ca7af9f492a226": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "0e1ead3dd9384c96b4737dddee823835": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_c2559631cd3641b88a5e5fe31d844211",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_a16bb1bc261a4660a7bbe443d1909dc4",
              "IPY_MODEL_d0673239579f4d9ca3465d8f0bf24d18"
            ]
          }
        },
        "c2559631cd3641b88a5e5fe31d844211": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "a16bb1bc261a4660a7bbe443d1909dc4": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_4d9a27c6b6b440098e4ca054927dec6e",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 650,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 650,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_eb6bb38c08d84f3386e5398fea1e32d8"
          }
        },
        "d0673239579f4d9ca3465d8f0bf24d18": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_27030226d32a47b3b102bbb3be87e3d1",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 650/650 [1:21:09&lt;00:00,  7.49s/it]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_eb448bd825e9465ca3f8c1e0df1d4f1a"
          }
        },
        "4d9a27c6b6b440098e4ca054927dec6e": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "eb6bb38c08d84f3386e5398fea1e32d8": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "27030226d32a47b3b102bbb3be87e3d1": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "eb448bd825e9465ca3f8c1e0df1d4f1a": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "bc0753880d174e6a96904f77a803853a": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_50da805cb4e04e3998b0fa4a9c7f5149",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_ed0e5b81c3434ad78594b72a0e717c35",
              "IPY_MODEL_f9fdf9ed465945089d1e99136c13d0fd"
            ]
          }
        },
        "50da805cb4e04e3998b0fa4a9c7f5149": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "ed0e5b81c3434ad78594b72a0e717c35": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_fbf412eafe6248c6a8a305044ef54763",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 650,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 650,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_7154ad28e32d4d719b17fd3e2fcc1a2a"
          }
        },
        "f9fdf9ed465945089d1e99136c13d0fd": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_3f793c468121465fbbe95515a9ac190c",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 650/650 [1:14:06&lt;00:00,  6.84s/it]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_37b9a0901be143c69043b234dec7b494"
          }
        },
        "fbf412eafe6248c6a8a305044ef54763": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "7154ad28e32d4d719b17fd3e2fcc1a2a": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "3f793c468121465fbbe95515a9ac190c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "37b9a0901be143c69043b234dec7b494": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "fe478569668e4040a9ee5b0f18114114": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_3087232fc1a242c7994d025d562b3c15",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_2fdfe7c2ff744bb48830fb4320dcff10",
              "IPY_MODEL_83cdd1d27d01462d8e7c3a5c9a879bb1"
            ]
          }
        },
        "3087232fc1a242c7994d025d562b3c15": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "2fdfe7c2ff744bb48830fb4320dcff10": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_5c43425620bf476e86abc419cea8c653",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 650,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 650,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_b3c767934da942918733432abbefa3b1"
          }
        },
        "83cdd1d27d01462d8e7c3a5c9a879bb1": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_6a85d51776844292a5d61d9ee79a648d",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 650/650 [1:07:12&lt;00:00,  6.20s/it]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_c108ed4ce4864cbc92899d29a891d252"
          }
        },
        "5c43425620bf476e86abc419cea8c653": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "b3c767934da942918733432abbefa3b1": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "6a85d51776844292a5d61d9ee79a648d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "c108ed4ce4864cbc92899d29a891d252": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        }
      }
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jcSSTuX5Ouzh",
        "colab_type": "text"
      },
      "source": [
        "![banner](https://drive.google.com/uc?id=1qLz2QP1w_46OSAHCqnrhY1kvACkkwCe4)\n",
        "\n",
        "**Desarrollado por:**\n",
        "\n",
        "- Eduard Alfonso Caballero Ponguta - 2151099\n",
        "- María Camila Aparicio Nova - 2161907\n",
        "- Iván Rodrigo Castillo Cañas - 2160023\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "enDx8olLtad6",
        "colab_type": "code",
        "cellView": "form",
        "colab": {}
      },
      "source": [
        "#@title Importar Librerías\n",
        "import pandas as pd\n",
        "import random\n",
        "import os\n",
        "import numpy as np\n",
        "import io\n",
        "import IPython\n",
        "from google.colab import output\n",
        "from datetime import datetime\n",
        "!pip install ffmpeg-python\n",
        "import ffmpeg\n",
        "\n",
        "from scipy.io import wavfile\n",
        "from scipy.fftpack import fft\n",
        "from scipy import signal\n",
        "from scipy.io.wavfile import read as wav_read\n",
        "import scipy\n",
        "\n",
        "\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "from sklearn.model_selection import train_test_split \n",
        "from sklearn.metrics import confusion_matrix\n",
        "\n",
        "from keras.utils import to_categorical\n",
        "import tensorflow as tf\n",
        "\n",
        "\n",
        "import librosa\n",
        "import librosa.display\n",
        "\n",
        "from tqdm.notebook import tqdm\n",
        "from IPython.display import clear_output\n",
        "from google.colab import drive\n",
        "import matplotlib.pyplot as plt\n",
        "from os.path import isdir, join\n",
        "from IPython.display import HTML, Audio\n",
        "from google.colab.output import eval_js\n",
        "from base64 import b64decode\n",
        "\n",
        "from keras.layers import Dense, Dropout, Flatten, Conv1D, Input, MaxPooling1D\n",
        "from keras.models import Model\n",
        "from keras.callbacks import EarlyStopping, ModelCheckpoint\n",
        "from keras import backend as K\n",
        "\n",
        "#Implementación\n",
        "import time\n",
        "!pip install pydub\n",
        "from pydub import AudioSegment\n",
        "from pydub.silence import split_on_silence\n",
        "\n",
        "%matplotlib inline\n",
        "clear_output()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bA4dvhOnKLyT",
        "colab_type": "code",
        "cellView": "form",
        "colab": {}
      },
      "source": [
        "#@title Cargar google drive\n",
        "drive.mount('/content/drive')\n",
        "os.chdir('/content/drive/My Drive/IA2/DATA/')\n",
        "clear_output()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VIajRloezv0b",
        "colab_type": "code",
        "cellView": "form",
        "colab": {}
      },
      "source": [
        "#@title Funciones\n",
        "AUDIO_HTML = \"\"\"\n",
        "<script>\n",
        "var my_div = document.createElement(\"DIV\");\n",
        "var screen = document.createElement(\"DIV\");\n",
        "var my_btn = document.createElement(\"BUTTON\");\n",
        "var t = document.createTextNode(\"Grabar\");\n",
        "var t2 = document.createTextNode(\"00 : 00\");\n",
        "\n",
        "screen.appendChild(t2);\n",
        "my_btn.appendChild(t);\n",
        "my_div.appendChild(my_btn);\n",
        "document.body.appendChild(screen);\n",
        "document.body.appendChild(my_div);\n",
        "\n",
        "var base64data = 0;\n",
        "var reader;\n",
        "var play = false;\n",
        "var recorder, gumStream;\n",
        "var recordButton = my_btn;\n",
        "var screenInfo = screen;\n",
        "var isMarch = false; \n",
        "var acumularTime = 0; \n",
        "\n",
        "function cronometro () { \n",
        "  timeActual = new Date();\n",
        "  acumularTime = timeActual - timeInicial;\n",
        "  acumularTime2 = new Date();\n",
        "  acumularTime2.setTime(acumularTime); \n",
        "  cc = Math.round(acumularTime2.getMilliseconds()/10);\n",
        "  ss = acumularTime2.getSeconds();\n",
        "  if (cc < 10) {cc = \"0\"+cc;}\n",
        "  if (ss < 10) {ss = \"0\"+ss;} \n",
        "  screenInfo.innerHTML = ss+\" : \"+cc;\n",
        "}\n",
        "\n",
        "function start() {\n",
        "  if (isMarch == false) { \n",
        "    timeInicial = new Date();\n",
        "    control = setInterval(cronometro,10);\n",
        "    isMarch = true;\n",
        "  }\n",
        "}\n",
        "\n",
        "function stop() { \n",
        "  if (isMarch == true) {\n",
        "    clearInterval(control);\n",
        "    isMarch = false;\n",
        "  }     \n",
        "}      \n",
        "\n",
        "\n",
        "var handleSuccess = function(stream) {\n",
        "  gumStream = stream;\n",
        "  var options = {\n",
        "    //bitsPerSecond: 8000, //chrome seems to ignore, always 48k\n",
        "    mimeType : 'audio/webm;codecs=opus'\n",
        "    //mimeType : 'audio/webm;codecs=pcm'\n",
        "  };            \n",
        "  recorder = new MediaRecorder(stream);\n",
        "  recorder.ondataavailable = function(e) {            \n",
        "    var url = URL.createObjectURL(e.data);\n",
        "    var preview = document.createElement('audio');\n",
        "    preview.controls = true;\n",
        "    preview.src = url;\n",
        "    document.body.appendChild(preview);\n",
        "\n",
        "    reader = new FileReader();\n",
        "    reader.readAsDataURL(e.data); \n",
        "    reader.onloadend = function() {\n",
        "      base64data = reader.result;\n",
        "    }\n",
        "  };\n",
        "  recorder.start();\n",
        "};\n",
        "\n",
        "function grabar() {\n",
        "  recordButton.innerText = \"Parar\";\n",
        "  navigator.mediaDevices.getUserMedia({audio: true}).then(handleSuccess);\n",
        "  start();\n",
        "}\n",
        "\n",
        "\n",
        "function toggleRecording() {\n",
        "  if (recorder && recorder.state == \"recording\") {\n",
        "      recorder.stop();\n",
        "      stop();\n",
        "      gumStream.getAudioTracks()[0].stop();\n",
        "      recordButton.innerText = \"Guardando el audio...\"\n",
        "  }\n",
        "}\n",
        "\n",
        "function sleep(ms) {\n",
        "  return new Promise(resolve => setTimeout(resolve, ms));\n",
        "}\n",
        "\n",
        "var data = new Promise(resolve=>{\n",
        "//recordButton.addEventListener(\"click\", toggleRecording);\n",
        "recordButton.onclick = ()=>{\n",
        "  console.log(\"click\")\n",
        "if (play){\n",
        "  console.log(\"stop\")\n",
        "  toggleRecording()\n",
        "}else{\n",
        "  console.log(\"play\")\n",
        "  grabar()\n",
        "  play = true\n",
        "}\n",
        "\n",
        "sleep(5000).then(() => {\n",
        "  resolve(base64data.toString())\n",
        "  recordButton.innerText = \"Guardado con exito!\"\n",
        "});\n",
        "\n",
        "}\n",
        "});\n",
        "      \n",
        "</script>\n",
        "\"\"\"\n",
        "BULB_HTML = \"\"\"\n",
        "<script>\n",
        "var my_div = document.createElement(\"DIV\");\n",
        "var image = document.createElement(\"IMG\");\n",
        "\n",
        "image.setAttribute(\"src\", \"https://media.geeksforgeeks.org/wp-content/uploads/OFFbulb.jpg\");\n",
        "document.body.appendChild(my_div);\n",
        "\n",
        "var div_image = image;\n",
        "\n",
        "function turnOnOff()  \n",
        "// taking image in image variable \n",
        "var image = document.getElementById('Image'); \n",
        "\n",
        "//Match the image name \n",
        "//whether it is ONbulb or OFFbulb \n",
        "//change image to OFFbulb.jpg if \n",
        "//it match with ONbulb otherwise \n",
        "//change it to ONbulb.jpg --> \n",
        "if (image.src.match(\"ONbulb\")) \n",
        "    div_image.src = \"https://media.geeksforgeeks.org/ \n",
        "                wp-content/uploads/OFFbulb.jpg\"; \n",
        "else\n",
        "    div_image.src = \"https://media.geeksforgeeks.org/ \n",
        "                wp-content/uploads/ONbulb.jpg\"; \n",
        "}\n",
        "</script>\n",
        "\"\"\"\n",
        "\n",
        "def log_specgram(audio, sample_rate, window_size=20,\n",
        "                 step_size=10, eps=1e-10):\n",
        "    nperseg = int(round(window_size * sample_rate / 1e3))\n",
        "    noverlap = int(round(step_size * sample_rate / 1e3))\n",
        "    freqs, times, spec = signal.spectrogram(audio,\n",
        "                                    fs=sample_rate,\n",
        "                                    window='hann',\n",
        "                                    nperseg=nperseg,\n",
        "                                    noverlap=noverlap,\n",
        "                                    detrend=False)\n",
        "    return freqs, times, np.log(spec.T.astype(np.float32) + eps)\n",
        "\n",
        "def extract_features(train_audio_path, file_name):\n",
        "  audio, sample_rate = librosa.load(train_audio_path+file_name, res_type='kaiser_fast') \n",
        "  mfccs = librosa.feature.mfcc(y=audio, sr=sample_rate, n_mfcc=80)\n",
        "  mfccs_processed = np.mean(mfccs.T,axis=0)\n",
        "  return mfccs_processed\n",
        "\n",
        "def extract_features_chroma(train_audio_path, file_name):\n",
        "  audio, sample_rate = librosa.load(train_audio_path+file_name, res_type='kaiser_fast') \n",
        "  stft = np.abs(librosa.stft(audio))\n",
        "  chroma = np.mean(librosa.feature.chroma_stft(S=stft, sr=sample_rate).T,axis=0)\n",
        "  return chroma\n",
        "\n",
        "def remove_label_from_file(label, fname):\n",
        "    return fname[len(label)+1:]\n",
        "\n",
        "def get_audio():\n",
        "  display(HTML(AUDIO_HTML))\n",
        "  data = eval_js(\"data\")\n",
        "  binary = b64decode(data.split(',')[1])\n",
        "  \n",
        "  process = (ffmpeg\n",
        "    .input('pipe:0')\n",
        "    .output('pipe:1', format='wav')\n",
        "    .run_async(pipe_stdin=True, pipe_stdout=True, pipe_stderr=True, quiet=True, overwrite_output=True)\n",
        "  )\n",
        "  output, err = process.communicate(input=binary)\n",
        "  \n",
        "  riff_chunk_size = len(output) - 8\n",
        "  # Break up the chunk size into four bytes, held in b.\n",
        "  q = riff_chunk_size\n",
        "  b = []\n",
        "  for i in range(4):\n",
        "      q, r = divmod(q, 256)\n",
        "      b.append(r)\n",
        "\n",
        "  # Replace bytes 4:8 in proc.stdout with the actual size of the RIFF chunk.\n",
        "  riff = output[:4] + bytes(b) + output[8:]\n",
        "\n",
        "  sr, audio = wav_read(io.BytesIO(riff))\n",
        "  now = datetime.now()\n",
        "  name = now.strftime('%d_%m_%Y_%H_%M_%S')\n",
        "  scipy.io.wavfile.write('/content/drive/My Drive/IA2/DATA/Recs/rec_'+name+'.wav', sr, audio)\n",
        "\n",
        "  return audio, sr, 'rec_'+name+'.wav'"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fTYcGaAWA8Vp",
        "colab_type": "text"
      },
      "source": [
        "##1. OBJETIVOS DEL PROYECTO"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LdtAnOZEO7Q5",
        "colab_type": "text"
      },
      "source": [
        "La accesibilidad de los dispositivos inteligentes en la actualidad es un factor incidente en la autonomía de tareas y flujo de actividades cotidianas. Existen asistentes por voz que se basan en tecnología con Inteligencia Artificial donde las redes neuronales han mostrado ser una herramienta versátil capaz de modelar todos los aspectos acústicos, fonéticos y lingüísticos asociados con la tarea de reconocimiento de voz.\n",
        "\n",
        "\n",
        "### **Objetivo general**\n",
        "\n",
        "- Reconocer comandos de voz simples en inglés mediante una interfaz que registra la voz y ejecuta acciones de acuerdo los comandos detectados en audios grabados, facilitando la interacción y accesibilidad con dispositivos inteligentes.\n",
        "\n",
        "### **Objetivos especificos**\n",
        "\n",
        "-  Construir modelos DNN, RNN, LSTM y GRU que permitan el reconocimiento de diferentes tipos de comandos de voz.\n",
        "- Evalúar los modelos entrenados mediante dos tipos de pruebas: por pablabras y por frases.\n",
        "-  Determinar el modelo que ofrezca mejores resultados en rendimiento y accuracy."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "e0RpODeTPAxi",
        "colab_type": "text"
      },
      "source": [
        "### Preguntas por responder\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "V44KC_HpPZ7I",
        "colab_type": "text"
      },
      "source": [
        "- **Problema a resolver:**  \n",
        "R: ¿Es posible reconocer comandos de voz simples que permitan ejecutar acciones en dispositivos con acceso a micrófono?\n",
        "- **¿Como se va resolver?**  \n",
        "R: Por medio de varios tratamientos que permitan una mejor representacion de la información y la construccion de modelos Deep Learning para reconocer los comandos hablados.\n",
        "- **¿Por qué se necesita resolver este problema?**  \n",
        "R: Porque el internet de las cosas (IoT) cada vez está más presente en los dispositivos y electrodomésticos que se usan diariamente, los cuales se reinventan para ser cada vez más compactos como, por ejemplo, poder *mandar* a un dispositivo mediante comando de voz para que ejecute tareas, sin la necesidad de entrar en contacto directo con una pantalla como es el caso de Alexa, Siri o AG.\n",
        "- **¿Para que?**  \n",
        "R: Reducir el uso de pantallas y facilitar el uso de dispositivos en personas con alguna discapacidad motriz.\n",
        "- **¿Para quien(es) seria util esta solucion?**  \n",
        "R: Para empresas desarrolladoras de dispositivos electronicos o servicios web, que requieran el uso de reconocimiento de voz."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_HlQ1bntQy84",
        "colab_type": "text"
      },
      "source": [
        "# 2. TRATAMIENTO DE AUDIOS\n",
        "El dataset que se utilizó para este proyecto se obtuvo de [kaggle](https://www.kaggle.com/c/tensorflow-speech-recognition-challenge ), el cual consta de las siguientes características:\n",
        "\n",
        "* 65.000 audios de comandos grabados en inglés por diferentes tonos de voz.\n",
        "* 30 clases (comandos).\n",
        "* Cada audio dura un segundo.\n",
        "* formato .wav\n",
        "\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yIaoP-HmpHkO",
        "colab_type": "text"
      },
      "source": [
        "### Selección de comandos\n",
        "Se seleccionan las grabaciones correspondientes de los comandos correspondientes a **yes, no, on, off, stop.**\n",
        "\n",
        "Además, se crea una clase llamada **unknown** a partir de la concatenación de los comandos one, five y dog. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "y-LT3hG2pQVk",
        "colab_type": "text"
      },
      "source": [
        "\n",
        "| Comandos | Descripción |\n",
        "|-----------------------|---------------------------------------------------------------------------------|\n",
        "| **yes** | 2377 audios de 1 segundo pronunciados por diferentes personas (formato: wav) |\n",
        "| **no** | 2375 audios de 1 segundo pronunciados por diferentes personas (formato: wav) |\n",
        "| **on** | 2365 audios de 1 segundo pronunciados por diferentes personas (formato: wav) |\n",
        "|  **off** | 2357 audios de 1 segundo pronunciados por diferentes personas (formato: wav) |\n",
        "|  **stop** | 2357 audios de 1 segundo pronunciados por diferentes personas (formato: wav) |\n",
        "|  **unknown** | 7021 audios de 1 segundo pronunciados por diferentes personas (formato: wav) |"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "j32zi71svig_",
        "colab_type": "code",
        "cellView": "form",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 524
        },
        "outputId": "79e6b2a2-1e40-49d9-d55a-24ef44d127a2"
      },
      "source": [
        "#@title Representación gráfica\n",
        "#@markdown Se grafica la representacion temporal y el espectograma de mel para una muestra  de cada clase\n",
        "train_audio_path = 'train/audio/'\n",
        "yes = '/yes/0a7c2a8d_nohash_0.wav'\n",
        "no = '/no/b4aa9fef_nohash_0.wav'\n",
        "on = '/on/5af0ca83_nohash_1.wav'\n",
        "off = '/off/924c1a04_nohash_1.wav'\n",
        "stop = '/stop/0fa1e7a9_nohash_0.wav'\n",
        "one = '/one/ffd2ba2f_nohash_2.wav'\n",
        "five = '/five/ff21fb59_nohash_1.wav'\n",
        "dog=  '/dog/fde2dee7_nohash_1.wav'\n",
        "\n",
        "samples_yes, sample_rate_yes = librosa.load(str(train_audio_path)+yes)\n",
        "S_yes = librosa.feature.melspectrogram(samples_yes, sr=sample_rate_yes, n_mels=128)\n",
        "log_S_yes = librosa.power_to_db(S_yes, ref=np.max)\n",
        "\n",
        "samples_no, sample_rate_no = librosa.load(str(train_audio_path)+no)\n",
        "S_no = librosa.feature.melspectrogram(samples_no, sr=sample_rate_no, n_mels=128)\n",
        "log_S_no = librosa.power_to_db(S_no, ref=np.max)\n",
        "\n",
        "samples_on, sample_rate_on = librosa.load(str(train_audio_path)+on)\n",
        "S_on = librosa.feature.melspectrogram(samples_on, sr=sample_rate_on, n_mels=128)\n",
        "log_S_on = librosa.power_to_db(S_on, ref=np.max)\n",
        "\n",
        "samples_off, sample_rate_off = librosa.load(str(train_audio_path)+off)\n",
        "S_off = librosa.feature.melspectrogram(samples_off, sr=sample_rate_off, n_mels=128)\n",
        "log_S_off = librosa.power_to_db(S_off, ref=np.max)\n",
        "\n",
        "samples_stop, sample_rate_stop = librosa.load(str(train_audio_path)+stop)\n",
        "S_stop = librosa.feature.melspectrogram(samples_stop, sr=sample_rate_stop, n_mels=128)\n",
        "log_S_stop = librosa.power_to_db(S_stop, ref=np.max)\n",
        "\n",
        "samples_one, sample_rate_one = librosa.load(str(train_audio_path)+one)\n",
        "S_one = librosa.feature.melspectrogram(samples_one, sr=sample_rate_one, n_mels=128)\n",
        "log_S_one = librosa.power_to_db(S_one, ref=np.max)\n",
        "\n",
        "samples_five, sample_rate_five = librosa.load(str(train_audio_path)+five)\n",
        "S_five = librosa.feature.melspectrogram(samples_five, sr=sample_rate_five, n_mels=128)\n",
        "log_S_five = librosa.power_to_db(S_five, ref=np.max)\n",
        "\n",
        "samples_dog, sample_rate_dog = librosa.load(str(train_audio_path)+dog)\n",
        "S_dog = librosa.feature.melspectrogram(samples_dog, sr=sample_rate_dog, n_mels=128)\n",
        "log_S_dog = librosa.power_to_db(S_dog, ref=np.max)\n",
        "\n",
        "#fig, [ax1, ax2, ax3, ax4, ax5] = plt.subplots(nrows=2, ncols=5)\n",
        "fig1 = plt.figure(figsize=(16, 4))\n",
        "\n",
        "plt.subplot(2,4,1)\n",
        "librosa.display.waveplot(samples_yes, sr=sample_rate_yes)\n",
        "plt.title('Representacion temporal \"yes\"')\n",
        "plt.ylabel('Amplitud')\n",
        "\n",
        "plt.subplot(2,4,5)\n",
        "librosa.display.specshow(log_S_yes, sr=sample_rate_yes, x_axis='time', y_axis='mel')\n",
        "plt.title('Espectograma de Mel \"yes\"')\n",
        "plt.colorbar(format='%+02.0f dB')\n",
        "plt.tight_layout()\n",
        "\n",
        "plt.subplot(2,4,2)\n",
        "librosa.display.waveplot(samples_no, sr=sample_rate_no)\n",
        "plt.title('Representacion temporal \"no\"')\n",
        "plt.ylabel('Amplitud')\n",
        "\n",
        "plt.subplot(2,4,6)\n",
        "librosa.display.specshow(log_S_no, sr=sample_rate_no, x_axis='time', y_axis='mel')\n",
        "plt.title('Espectograma de Mel \"no\"')\n",
        "plt.colorbar(format='%+02.0f dB')\n",
        "plt.tight_layout()\n",
        "\n",
        "plt.subplot(2,4,3)\n",
        "librosa.display.waveplot(samples_on, sr=sample_rate_on)\n",
        "plt.title('Representacion temporal \"on\"')\n",
        "plt.ylabel('Amplitud')\n",
        "\n",
        "plt.subplot(2,4,7)\n",
        "librosa.display.specshow(log_S_on, sr=sample_rate_on, x_axis='time', y_axis='mel')\n",
        "plt.title('Espectograma de Mel \"on\"')\n",
        "plt.colorbar(format='%+02.0f dB')\n",
        "plt.tight_layout()\n",
        "\n",
        "\n",
        "plt.subplot(2,4,4)\n",
        "librosa.display.waveplot(samples_off, sr=sample_rate_off)\n",
        "plt.title('Representacion temporal \"off\"')\n",
        "plt.ylabel('Amplitud')\n",
        "\n",
        "plt.subplot(2,4,8)\n",
        "librosa.display.specshow(log_S_off, sr=sample_rate_off, x_axis='time', y_axis='mel')\n",
        "plt.title('Espectograma de Mel \"off\"')\n",
        "plt.colorbar(format='%+02.0f dB')\n",
        "plt.tight_layout()\n",
        "\n",
        "fig2 = plt.figure(figsize=(16, 4))\n",
        "\n",
        "plt.subplot(2,4,1)\n",
        "librosa.display.waveplot(samples_stop, sr=sample_rate_stop)\n",
        "plt.title('Representacion temporal \"stop\"')\n",
        "plt.ylabel('Amplitud')\n",
        "\n",
        "plt.subplot(2,4,5)\n",
        "librosa.display.specshow(log_S_stop, sr=sample_rate_stop, x_axis='time', y_axis='mel')\n",
        "plt.title('Espectograma de Mel \"stop\"')\n",
        "plt.colorbar(format='%+02.0f dB')\n",
        "plt.tight_layout()\n",
        "\n",
        "plt.subplot(2,4,2)\n",
        "librosa.display.waveplot(samples_one, sr=sample_rate_one)\n",
        "plt.title('Representacion temporal \"one\"')\n",
        "plt.ylabel('Amplitud')\n",
        "\n",
        "plt.subplot(2,4,6)\n",
        "librosa.display.specshow(log_S_one, sr=sample_rate_one, x_axis='time', y_axis='mel')\n",
        "plt.title('Espectograma de Mel \"one\"')\n",
        "plt.colorbar(format='%+02.0f dB')\n",
        "plt.tight_layout()\n",
        "\n",
        "plt.subplot(2,4,3)\n",
        "librosa.display.waveplot(samples_five, sr=sample_rate_five)\n",
        "plt.title('Representacion temporal \"five\"')\n",
        "plt.ylabel('Amplitud')\n",
        "\n",
        "plt.subplot(2,4,7)\n",
        "librosa.display.specshow(log_S_five, sr=sample_rate_five, x_axis='time', y_axis='mel')\n",
        "plt.title('Espectograma de Mel \"five\"')\n",
        "plt.colorbar(format='%+02.0f dB')\n",
        "plt.tight_layout()\n",
        "\n",
        "plt.subplot(2,4,4)\n",
        "librosa.display.waveplot(samples_dog, sr=sample_rate_dog)\n",
        "plt.title('Representacion temporal \"dog\"')\n",
        "plt.ylabel('Amplitud')\n",
        "\n",
        "plt.subplot(2,4,8)\n",
        "librosa.display.specshow(log_S_dog, sr=sample_rate_dog, x_axis='time', y_axis='mel')\n",
        "plt.title('Espectograma de Mel \"dog\"')\n",
        "plt.colorbar(format='%+02.0f dB')\n",
        "plt.tight_layout()\n",
        "plt.show()\n",
        "# Audio(samples, rate=sample_rate)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1152x288 with 12 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1152x288 with 12 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IHOPfDFT_0zl",
        "colab_type": "text"
      },
      "source": [
        "### Dataset de entrenamiento\n",
        "Se extraen las características MFCC y Chroma con librosa para crear un dataset de entrenamiento. Además, se particiona este dataset en una cantidad de audios para entrenamiento y testeo respectivamente."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZKXbmE8eDw0k",
        "colab_type": "text"
      },
      "source": [
        "![alt text](https://drive.google.com/uc?id=1-99kFGwxhn7wR5DZRMuQ6kLFix48k0KS)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LpHWCMbu232K",
        "colab_type": "code",
        "cellView": "both",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 686,
          "referenced_widgets": [
            "23c1f6a1983944a8951eb86aaee87a97",
            "1a6e07770c564ab1921df9fc852b3e5b",
            "0ac2ac9530ec4eb29d30a0565761e691",
            "74b0ce2b527344b6830f54333a54f21e",
            "1c7f757b3ceb463f946ee4e1767713b5",
            "e32333d33f2346b2a2c7a1e5e2e9f57d",
            "5cc377511d7e4142a9321b6016b69942",
            "439299fc8fd84a9e85f67afe80d7d666",
            "965d954a066c4acebdbc404e5362a5c5",
            "37e475f586a5434db67ffae8c9646989",
            "2d3a9ce70da74f81812eddc03e4834c7",
            "73e1ddb9cb744e5b8754a435d2b41577",
            "8bd7fb2857114340a772fdab802005b3",
            "c94eef4b6a3d4af48ee4467bb578bc42",
            "5e9d515965d44c019205e5fd782a2326",
            "081a5140391047d5808e9fa83a209eee",
            "c31d3ce029084266a4240f8bd2d57430",
            "f0e93407f8414a51b0114f8e1a69d3f9",
            "dbcc3cb7088e48a2a7895da21bf244bd",
            "0cfc5bb596cf44cbaad01f09e647f748",
            "823dcaccc7904a61a32e034c9589a689",
            "afb64f031f2d4af3975309620b643185",
            "1738aa539ac842ec912041b0e5599c03",
            "d072fa81c5344451a9f422200ea2e5d2",
            "49d69c43b6ef4dff8ecec9f9469e104d",
            "d89ca8bcc8db4cb9826a9f3baecc9ad2",
            "9c417a0c2f0741cc87cc97f0edb1530f",
            "1c40d4b3feff4cbb922069ccee260ad0",
            "52b5d0b5c03b42ccb45777db7e726d40",
            "18b07cac759640f083c3aee38b98e249",
            "b9edb41cc3bb453ab5e377fc446c0fd1",
            "2b6bf280dbad4598adfb76baa563c3e2",
            "adbb91f1450847faa861d44ed598c204",
            "f6105d82b52843bba7f9c1c0021067b2",
            "bc2f1fbdf34e4435b4fc4b45b54b5d79",
            "00a58be961a2458498e34b4f3e6b5ea6",
            "c987d7bfe36046f6b9b8e8c7bf01bf3e",
            "b3d0c9cf18394831a7c921780531c6cc",
            "b05650179ffd4323b4441e3a48120ea6",
            "7f73c44b92714c5599ca7af9f492a226",
            "0e1ead3dd9384c96b4737dddee823835",
            "c2559631cd3641b88a5e5fe31d844211",
            "a16bb1bc261a4660a7bbe443d1909dc4",
            "d0673239579f4d9ca3465d8f0bf24d18",
            "4d9a27c6b6b440098e4ca054927dec6e",
            "eb6bb38c08d84f3386e5398fea1e32d8",
            "27030226d32a47b3b102bbb3be87e3d1",
            "eb448bd825e9465ca3f8c1e0df1d4f1a",
            "bc0753880d174e6a96904f77a803853a",
            "50da805cb4e04e3998b0fa4a9c7f5149",
            "ed0e5b81c3434ad78594b72a0e717c35",
            "f9fdf9ed465945089d1e99136c13d0fd",
            "fbf412eafe6248c6a8a305044ef54763",
            "7154ad28e32d4d719b17fd3e2fcc1a2a",
            "3f793c468121465fbbe95515a9ac190c",
            "37b9a0901be143c69043b234dec7b494",
            "fe478569668e4040a9ee5b0f18114114",
            "3087232fc1a242c7994d025d562b3c15",
            "2fdfe7c2ff744bb48830fb4320dcff10",
            "83cdd1d27d01462d8e7c3a5c9a879bb1",
            "5c43425620bf476e86abc419cea8c653",
            "b3c767934da942918733432abbefa3b1",
            "6a85d51776844292a5d61d9ee79a648d",
            "c108ed4ce4864cbc92899d29a891d252"
          ]
        },
        "outputId": "2387ce17-ad1c-4d46-e885-3a5c6a629f23"
      },
      "source": [
        "#@title Partición del dataset\n",
        "size = 650 #@param {type:\"slider\", min:50, max:2300, step:100}\n",
        "shuffle = \"True\" #@param ['True','False']\n",
        "# train_labels = os.listdir(train_audio_path)\n",
        "# train_labels.remove('_background_noise_')\n",
        "\n",
        "\n",
        "labels_to_keep = ['yes', 'no', 'on', 'off','stop', 'one','five','dog']\n",
        "# labels_to_keep = ['up', 'down', 'left', 'right']\n",
        "\n",
        "#num_labels = len(labels_to_keep)\n",
        "train_audio_path = 'train/audio/'\n",
        "\n",
        "# train_file_labels = dict()\n",
        "features = []\n",
        "unknow = []\n",
        "\n",
        "for label in labels_to_keep[:5]:\n",
        "  files = os.listdir(train_audio_path + '/' + label)\n",
        "  if (shuffle == \"True\"): \n",
        "    random.shuffle(files)\n",
        "  for f in tqdm(files[:size]):\n",
        "    file_name = label + '/' + f\n",
        "    mfcc  = extract_features(train_audio_path, str(file_name))\n",
        "    chroma  = extract_features_chroma(train_audio_path, str(file_name))\n",
        "    features.append([f, np.concatenate((mfcc, chroma)), label])\n",
        "\n",
        "#Ruido Class Unknow\n",
        "for label in labels_to_keep[5:]:\n",
        "  files = os.listdir(train_audio_path + '/' + label)\n",
        "  if (shuffle == \"True\"): \n",
        "    random.shuffle(files)\n",
        "  for f in tqdm(files[:size]):\n",
        "    file_name = label + '/' + f\n",
        "    mfcc  = extract_features(train_audio_path, str(file_name))\n",
        "    chroma  = extract_features_chroma(train_audio_path, str(file_name))\n",
        "    unknow.append([f, np.concatenate((mfcc, chroma)), 'unknow'])\n",
        "random.shuffle(unknow)\n",
        "features += unknow\n",
        "data_audio = pd.DataFrame(features, columns=['file','features','class_label'])\n",
        "data_audio.sample(5)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "23c1f6a1983944a8951eb86aaee87a97",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=650.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/librosa/core/pitch.py:145: UserWarning: Trying to estimate tuning from empty frequency set.\n",
            "  warnings.warn('Trying to estimate tuning from empty frequency set.')\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "965d954a066c4acebdbc404e5362a5c5",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=650.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "c31d3ce029084266a4240f8bd2d57430",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=650.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "49d69c43b6ef4dff8ecec9f9469e104d",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=650.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/librosa/core/pitch.py:145: UserWarning: Trying to estimate tuning from empty frequency set.\n",
            "  warnings.warn('Trying to estimate tuning from empty frequency set.')\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "adbb91f1450847faa861d44ed598c204",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=650.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "0e1ead3dd9384c96b4737dddee823835",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=650.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "bc0753880d174e6a96904f77a803853a",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=650.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "fe478569668e4040a9ee5b0f18114114",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=650.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/librosa/core/pitch.py:145: UserWarning: Trying to estimate tuning from empty frequency set.\n",
            "  warnings.warn('Trying to estimate tuning from empty frequency set.')\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>file</th>\n",
              "      <th>features</th>\n",
              "      <th>class_label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>3800</th>\n",
              "      <td>d1214f15_nohash_1.wav</td>\n",
              "      <td>[-567.0991045495317, 73.8521946474203, -28.526...</td>\n",
              "      <td>unknow</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>347</th>\n",
              "      <td>d90b4138_nohash_3.wav</td>\n",
              "      <td>[-644.4743070646825, 52.65493282993103, -1.445...</td>\n",
              "      <td>yes</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>686</th>\n",
              "      <td>573cdb8a_nohash_1.wav</td>\n",
              "      <td>[-679.8258806071126, 79.77227173624026, -19.74...</td>\n",
              "      <td>no</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4277</th>\n",
              "      <td>5b1db3ee_nohash_0.wav</td>\n",
              "      <td>[-548.0019147073128, 95.93727402425051, -18.87...</td>\n",
              "      <td>unknow</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2200</th>\n",
              "      <td>d0faf7e4_nohash_4.wav</td>\n",
              "      <td>[-395.0501611888109, 126.29257403457987, 19.88...</td>\n",
              "      <td>off</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                       file  ... class_label\n",
              "3800  d1214f15_nohash_1.wav  ...      unknow\n",
              "347   d90b4138_nohash_3.wav  ...         yes\n",
              "686   573cdb8a_nohash_1.wav  ...          no\n",
              "4277  5b1db3ee_nohash_0.wav  ...      unknow\n",
              "2200  d0faf7e4_nohash_4.wav  ...         off\n",
              "\n",
              "[5 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YpC-RvsVXgzw",
        "colab_type": "code",
        "cellView": "both",
        "colab": {}
      },
      "source": [
        "#@title Partición de test y codificación\n",
        "test_size = 0.25 #@param {type:\"slider\", min:0.1, max:1, step:0.05}\n",
        "\n",
        "# Convertimos las features y labels a numpy arrays\n",
        "num_labels = 6\n",
        "X = np.array(data_audio.features.tolist())\n",
        "y = np.array(data_audio.class_label.tolist())\n",
        "\n",
        "# Codificamos los labels con Encoder\n",
        "le = LabelEncoder()\n",
        "y_encode = to_categorical(le.fit_transform(y))\n",
        "\n",
        "# Split\n",
        "x_train, x_test, y_train, y_test = train_test_split(X, y_encode, test_size=test_size)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bbX6JZkubp5A",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "d1660466-2bbb-4712-cc9c-3240e59029ba"
      },
      "source": [
        "print(f\"Dataset shape: Train:{x_train.shape}, Test:{y_test.shape}\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Dataset shape: Train:(3900, 92), Test:(1300, 6)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gShNjdIyAcos",
        "colab_type": "text"
      },
      "source": [
        "#3. DEFINICIÓN DE MODELOS\n",
        "Se definen y entrenan modelos DNN y RNN con las características extraídas en el enunciado anterior."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tX6ajhg1P6jC",
        "colab_type": "text"
      },
      "source": [
        "##Entrenamiento Modelo RNN"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Bt8cMZL_P_rp",
        "colab_type": "code",
        "cellView": "both",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 391
        },
        "outputId": "10eb39b5-fda8-442c-a560-556388bd1398"
      },
      "source": [
        "tf.keras.backend.clear_session()\n",
        "model = tf.keras.models.Sequential( )\n",
        "simpleRNN1 = tf.keras.layers.Bidirectional((tf.keras.layers.SimpleRNN(units=512,\n",
        "                                       return_sequences=True,\n",
        "                                       activation='tanh',\n",
        "                                       name='RNN_layer1')))\n",
        "\n",
        "simpleRNN2 = tf.keras.layers.Bidirectional((tf.keras.layers.SimpleRNN(units=512,\n",
        "                                       return_sequences=True,\n",
        "                                       activation='tanh',\n",
        "                                       name='RNN_layer2')))\n",
        "\n",
        "simpleRNN3 = tf.keras.layers.Bidirectional((tf.keras.layers.SimpleRNN(units=512,\n",
        "                                       return_sequences=False,\n",
        "                                       activation='tanh',\n",
        "                                       name='RNN_layer3')))\n",
        "# To calculate the output value                                           \n",
        "dense1 = tf.keras.layers.Dense(units=256, \n",
        "                                   activation='relu',\n",
        "                                   name='dense1')\n",
        "dense2 = tf.keras.layers.Dense(units=128, \n",
        "                                   activation='relu',\n",
        "                                   name='dense2')\n",
        "dense3 = tf.keras.layers.Dense(units = num_labels,\n",
        "                                  activation='softmax',\n",
        "                                  name='ouput_layer')\n",
        "############################################################################\n",
        "# Create the model\n",
        "input_ = tf.keras.layers.Input((92,1))\n",
        "x = simpleRNN1(input_)\n",
        "x = simpleRNN2(x)\n",
        "x = simpleRNN3(x)\n",
        "x = dense1(x)\n",
        "x = dense2(x)\n",
        "y = dense3(x)\n",
        "############################################################################\n",
        "\n",
        "model = tf.keras.Model(inputs=input_, outputs=y, name=\"Modelo_RNN\")\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"Modelo_RNN\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_1 (InputLayer)         [(None, 92, 1)]           0         \n",
            "_________________________________________________________________\n",
            "bidirectional (Bidirectional (None, 92, 1024)          526336    \n",
            "_________________________________________________________________\n",
            "bidirectional_1 (Bidirection (None, 92, 1024)          1573888   \n",
            "_________________________________________________________________\n",
            "bidirectional_2 (Bidirection (None, 1024)              1573888   \n",
            "_________________________________________________________________\n",
            "dense1 (Dense)               (None, 256)               262400    \n",
            "_________________________________________________________________\n",
            "dense2 (Dense)               (None, 128)               32896     \n",
            "_________________________________________________________________\n",
            "ouput_layer (Dense)          (None, 6)                 774       \n",
            "=================================================================\n",
            "Total params: 3,970,182\n",
            "Trainable params: 3,970,182\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SrMuUFbmkFcM",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        },
        "outputId": "a206a419-f40d-40c7-dd62-dcb5bd503932"
      },
      "source": [
        "X = x_train.reshape(x_train.shape[0],x_train.shape[1],1)\n",
        "X.shape\n",
        "y_train.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(2925, 6)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 53
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "qeONKrQ2dK1_"
      },
      "source": [
        "### 50 Epochs"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "cellView": "both",
        "id": "HSnorHIRdK2E",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "382b5ff5-17c2-4472-e01f-8ba90d440a2e"
      },
      "source": [
        "#@title Entrenamiento\n",
        "epochs = 50 #@param {type:\"slider\", min:50, max:5000, step:50}\n",
        "\n",
        "optimizer = tf.keras.optimizers.SGD(lr=1e-2, momentum=0.9)\n",
        "# optimizer = tf.keras.optimizers.Adam(lr=1e-4)\n",
        "# optimizer = tf.keras.optimizers.Adadelta(lr=1e-4)\n",
        "\n",
        "# optimizer = 'adam' #@param [\"adam\", \"SGD\", \"adadelta\"]\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              metrics=['accuracy'],\n",
        "              optimizer=optimizer)\n",
        "X = x_train.reshape(x_train.shape[0],x_train.shape[1],1)\n",
        "# print(X.shape, y_train.shape)\n",
        "history = model.fit(X, y_train,epochs=epochs,verbose=1)\n",
        "\n",
        "acc = history.history['accuracy']\n",
        "loss = history.history['loss']\n",
        "\n",
        "epochs=range(len(loss)) # Get number of epochs\n",
        "\n",
        "#------------------------------------------------\n",
        "# Plot acc and Loss\n",
        "#------------------------------------------------\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.plot(epochs, acc, 'red', label='acc')\n",
        "plt.title('Acc')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.show();\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.title('Loss')\n",
        "plt.plot(epochs, loss, 'black', label='loss')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Error\")\n",
        "plt.show();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 1.7214 - accuracy: 0.3436\n",
            "Epoch 2/50\n",
            "122/122 [==============================] - 32s 259ms/step - loss: 1.5776 - accuracy: 0.3938\n",
            "Epoch 3/50\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 1.4186 - accuracy: 0.4495\n",
            "Epoch 4/50\n",
            "122/122 [==============================] - 32s 259ms/step - loss: 1.3911 - accuracy: 0.4633\n",
            "Epoch 5/50\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 1.3520 - accuracy: 0.4728\n",
            "Epoch 6/50\n",
            "122/122 [==============================] - 32s 261ms/step - loss: 1.3419 - accuracy: 0.4733\n",
            "Epoch 7/50\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 1.3554 - accuracy: 0.4654\n",
            "Epoch 8/50\n",
            "122/122 [==============================] - 32s 261ms/step - loss: 1.3174 - accuracy: 0.4900\n",
            "Epoch 9/50\n",
            "122/122 [==============================] - 32s 262ms/step - loss: 1.3095 - accuracy: 0.4921\n",
            "Epoch 10/50\n",
            "122/122 [==============================] - 32s 263ms/step - loss: 1.2861 - accuracy: 0.4992\n",
            "Epoch 11/50\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 1.2824 - accuracy: 0.5000\n",
            "Epoch 12/50\n",
            "122/122 [==============================] - 32s 258ms/step - loss: 1.2655 - accuracy: 0.5108\n",
            "Epoch 13/50\n",
            "122/122 [==============================] - 31s 256ms/step - loss: 1.2728 - accuracy: 0.4987\n",
            "Epoch 14/50\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 1.2461 - accuracy: 0.5218\n",
            "Epoch 15/50\n",
            "122/122 [==============================] - 31s 254ms/step - loss: 1.2343 - accuracy: 0.5185\n",
            "Epoch 16/50\n",
            "122/122 [==============================] - 31s 252ms/step - loss: 1.2304 - accuracy: 0.5187\n",
            "Epoch 17/50\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 1.2060 - accuracy: 0.5272\n",
            "Epoch 18/50\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 1.1875 - accuracy: 0.5331\n",
            "Epoch 19/50\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 1.1868 - accuracy: 0.5436\n",
            "Epoch 20/50\n",
            "122/122 [==============================] - 31s 256ms/step - loss: 1.2084 - accuracy: 0.5290\n",
            "Epoch 21/50\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 1.2299 - accuracy: 0.5169\n",
            "Epoch 22/50\n",
            "122/122 [==============================] - 31s 254ms/step - loss: 1.2413 - accuracy: 0.5200\n",
            "Epoch 23/50\n",
            "122/122 [==============================] - 31s 256ms/step - loss: 1.1910 - accuracy: 0.5410\n",
            "Epoch 24/50\n",
            "122/122 [==============================] - 31s 253ms/step - loss: 1.1653 - accuracy: 0.5482\n",
            "Epoch 25/50\n",
            "122/122 [==============================] - 32s 263ms/step - loss: 1.2070 - accuracy: 0.5331\n",
            "Epoch 26/50\n",
            "122/122 [==============================] - 32s 264ms/step - loss: 1.2173 - accuracy: 0.5213\n",
            "Epoch 27/50\n",
            "122/122 [==============================] - 33s 270ms/step - loss: 1.2016 - accuracy: 0.5328\n",
            "Epoch 28/50\n",
            "122/122 [==============================] - 33s 274ms/step - loss: 1.1662 - accuracy: 0.5469\n",
            "Epoch 29/50\n",
            "122/122 [==============================] - 33s 271ms/step - loss: 1.1385 - accuracy: 0.5564\n",
            "Epoch 30/50\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 1.1349 - accuracy: 0.5521\n",
            "Epoch 31/50\n",
            "122/122 [==============================] - 32s 262ms/step - loss: 1.1026 - accuracy: 0.5738\n",
            "Epoch 32/50\n",
            "122/122 [==============================] - 32s 262ms/step - loss: 1.0981 - accuracy: 0.5777\n",
            "Epoch 33/50\n",
            "122/122 [==============================] - 32s 262ms/step - loss: 1.1269 - accuracy: 0.5782\n",
            "Epoch 34/50\n",
            "122/122 [==============================] - 32s 264ms/step - loss: 1.0786 - accuracy: 0.5856\n",
            "Epoch 35/50\n",
            "122/122 [==============================] - 32s 259ms/step - loss: 1.0831 - accuracy: 0.5885\n",
            "Epoch 36/50\n",
            "122/122 [==============================] - 32s 264ms/step - loss: 1.0772 - accuracy: 0.5879\n",
            "Epoch 37/50\n",
            "122/122 [==============================] - 31s 258ms/step - loss: 1.0181 - accuracy: 0.6164\n",
            "Epoch 38/50\n",
            "122/122 [==============================] - 32s 259ms/step - loss: 0.9880 - accuracy: 0.6187\n",
            "Epoch 39/50\n",
            "122/122 [==============================] - 32s 261ms/step - loss: 1.0119 - accuracy: 0.6092\n",
            "Epoch 40/50\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 0.9987 - accuracy: 0.6062\n",
            "Epoch 41/50\n",
            "122/122 [==============================] - 31s 254ms/step - loss: 0.9976 - accuracy: 0.6167\n",
            "Epoch 42/50\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 1.0298 - accuracy: 0.6041\n",
            "Epoch 43/50\n",
            "122/122 [==============================] - 31s 254ms/step - loss: 0.9676 - accuracy: 0.6377\n",
            "Epoch 44/50\n",
            "122/122 [==============================] - 30s 249ms/step - loss: 1.0067 - accuracy: 0.6195\n",
            "Epoch 45/50\n",
            "122/122 [==============================] - 30s 246ms/step - loss: 0.9277 - accuracy: 0.6464\n",
            "Epoch 46/50\n",
            "122/122 [==============================] - 30s 245ms/step - loss: 0.8902 - accuracy: 0.6554\n",
            "Epoch 47/50\n",
            "122/122 [==============================] - 30s 249ms/step - loss: 0.9146 - accuracy: 0.6533\n",
            "Epoch 48/50\n",
            "122/122 [==============================] - 30s 248ms/step - loss: 0.8892 - accuracy: 0.6633\n",
            "Epoch 49/50\n",
            "122/122 [==============================] - 31s 253ms/step - loss: 0.8469 - accuracy: 0.6831\n",
            "Epoch 50/50\n",
            "122/122 [==============================] - 30s 246ms/step - loss: 0.8263 - accuracy: 0.6885\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "cellView": "both",
        "id": "fHW_ia8XdK2R",
        "colab": {}
      },
      "source": [
        "#@markdown optimizer_epochs_samples_features_accuracy_version\n",
        "\n",
        "nombre = \"RNN_SGD_50e_650m_2f_0.76_V1\" #@param {type:\"string\"}\n",
        "model.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.h5')\n",
        "np.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.npy', le.classes_)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "lQ_Smw8r8JQz",
        "colab": {}
      },
      "source": [
        "# Cargar modelo\n",
        "model = tf.keras.models.load_model('/content/drive/My Drive/IA2/DATA/Models/model_RNN_SGD_50e_650m_2f_0.76_V1.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "LjgKzUMz8JQ8",
        "cellView": "both",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "78b1c922-c025-475a-b223-434c9ff86d83"
      },
      "source": [
        "#@title Predecición\n",
        "X_test = x_test.reshape(x_test.shape[0],x_test.shape[1],1)\n",
        "y_predict = np.argmax(model.predict(X_test), axis=1)\n",
        "y_test_single = np.argmax(y_test, axis=1)\n",
        "conf_matrix = confusion_matrix(y_test_single, y_predict)\n",
        "print(conf_matrix)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[ 85   4  13  14  18   7]\n",
            " [ 14 104  25  26  13   6]\n",
            " [ 25  27  69  18  26   3]\n",
            " [ 10  19   9  95  18  11]\n",
            " [124  42  72  85 134  31]\n",
            " [  9   3   3  15  10 113]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "kGxs3J1x8JRA",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "d8e5716f-5591-4fe0-88b1-9ec61b079e9a"
      },
      "source": [
        "text = [\"ON\", \"OFF\", \"YES\", \"NO\", \"STOP\", \"UNKNOWN\"]\n",
        "#Vector para gráficas\n",
        "RNN50=[]\n",
        "for i in range(len(text)):\n",
        "  acierto_promedio = conf_matrix[i][i]/(np.sum(conf_matrix[i], axis=0))\n",
        "  RNN50.append(np.round(acierto_promedio, 2))\n",
        "  print(f\"Aciertos clase {text[i]}: {np.round(acierto_promedio*100, 2)}%\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Aciertos clase ON: 60.28%\n",
            "Aciertos clase OFF: 55.32%\n",
            "Aciertos clase YES: 41.07%\n",
            "Aciertos clase NO: 58.64%\n",
            "Aciertos clase STOP: 27.46%\n",
            "Aciertos clase UNKNOWN: 73.86%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZP6xAjQ-cswM",
        "colab_type": "text"
      },
      "source": [
        "### 100 Epochs "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FtlN-TArQARy",
        "colab_type": "code",
        "cellView": "both",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "f8c47cb6-0acd-4845-bbb9-2ededf991576"
      },
      "source": [
        "#@title Entrenamiento\n",
        "epochs = 100 #@param {type:\"slider\", min:50, max:5000, step:50}\n",
        "\n",
        "optimizer = tf.keras.optimizers.SGD(lr=1e-2, momentum=0.9)\n",
        "# optimizer = tf.keras.optimizers.Adam(lr=1e-4)\n",
        "# optimizer = tf.keras.optimizers.Adadelta(lr=1e-4)\n",
        "\n",
        "# optimizer = 'adam' #@param [\"adam\", \"SGD\", \"adadelta\"]\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              metrics=['accuracy'],\n",
        "              optimizer=optimizer)\n",
        "X = x_train.reshape(x_train.shape[0],x_train.shape[1],1)\n",
        "# print(X.shape, y_train.shape)\n",
        "history = model.fit(X, y_train,epochs=epochs,verbose=1)\n",
        "\n",
        "acc = history.history['accuracy']\n",
        "loss = history.history['loss']\n",
        "\n",
        "epochs=range(len(loss)) # Get number of epochs\n",
        "\n",
        "#------------------------------------------------\n",
        "# Plot acc and Loss\n",
        "#------------------------------------------------\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.plot(epochs, acc, 'red', label='acc')\n",
        "plt.title('Acc')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.show();\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.title('Loss')\n",
        "plt.plot(epochs, loss, 'black', label='loss')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Error\")\n",
        "plt.show();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "122/122 [==============================] - 33s 267ms/step - loss: 1.2876 - accuracy: 0.5085\n",
            "Epoch 2/100\n",
            "122/122 [==============================] - 33s 267ms/step - loss: 1.1208 - accuracy: 0.5710\n",
            "Epoch 3/100\n",
            "122/122 [==============================] - 32s 266ms/step - loss: 1.0425 - accuracy: 0.5928\n",
            "Epoch 4/100\n",
            "122/122 [==============================] - 32s 265ms/step - loss: 1.0609 - accuracy: 0.5926\n",
            "Epoch 5/100\n",
            "122/122 [==============================] - 32s 265ms/step - loss: 1.1061 - accuracy: 0.5695\n",
            "Epoch 6/100\n",
            "122/122 [==============================] - 33s 271ms/step - loss: 1.1501 - accuracy: 0.5503\n",
            "Epoch 7/100\n",
            "122/122 [==============================] - 32s 265ms/step - loss: 1.0840 - accuracy: 0.5697\n",
            "Epoch 8/100\n",
            "122/122 [==============================] - 33s 267ms/step - loss: 1.0597 - accuracy: 0.5885\n",
            "Epoch 9/100\n",
            "122/122 [==============================] - 32s 266ms/step - loss: 1.0284 - accuracy: 0.5964\n",
            "Epoch 10/100\n",
            "122/122 [==============================] - 32s 265ms/step - loss: 0.9741 - accuracy: 0.6133\n",
            "Epoch 11/100\n",
            "122/122 [==============================] - 32s 266ms/step - loss: 0.9439 - accuracy: 0.6238\n",
            "Epoch 12/100\n",
            "122/122 [==============================] - 32s 264ms/step - loss: 0.8860 - accuracy: 0.6544\n",
            "Epoch 13/100\n",
            "122/122 [==============================] - 32s 264ms/step - loss: 0.8686 - accuracy: 0.6590\n",
            "Epoch 14/100\n",
            "122/122 [==============================] - 32s 265ms/step - loss: 0.8345 - accuracy: 0.6774\n",
            "Epoch 15/100\n",
            "122/122 [==============================] - 33s 269ms/step - loss: 0.8732 - accuracy: 0.6651\n",
            "Epoch 16/100\n",
            "122/122 [==============================] - 32s 263ms/step - loss: 0.8230 - accuracy: 0.6815\n",
            "Epoch 17/100\n",
            "122/122 [==============================] - 32s 265ms/step - loss: 0.7907 - accuracy: 0.6910\n",
            "Epoch 18/100\n",
            "122/122 [==============================] - 32s 263ms/step - loss: 0.7607 - accuracy: 0.7115\n",
            "Epoch 19/100\n",
            "122/122 [==============================] - 32s 264ms/step - loss: 0.8226 - accuracy: 0.6833\n",
            "Epoch 20/100\n",
            "122/122 [==============================] - 32s 263ms/step - loss: 0.8895 - accuracy: 0.6526\n",
            "Epoch 21/100\n",
            "122/122 [==============================] - 32s 262ms/step - loss: 0.8338 - accuracy: 0.6736\n",
            "Epoch 22/100\n",
            "122/122 [==============================] - 32s 262ms/step - loss: 0.7686 - accuracy: 0.7044\n",
            "Epoch 23/100\n",
            "122/122 [==============================] - 32s 263ms/step - loss: 0.7218 - accuracy: 0.7244\n",
            "Epoch 24/100\n",
            "122/122 [==============================] - 32s 263ms/step - loss: 0.7000 - accuracy: 0.7279\n",
            "Epoch 25/100\n",
            "122/122 [==============================] - 33s 270ms/step - loss: 0.7273 - accuracy: 0.7177\n",
            "Epoch 26/100\n",
            "122/122 [==============================] - 34s 278ms/step - loss: 0.7447 - accuracy: 0.7085\n",
            "Epoch 27/100\n",
            "122/122 [==============================] - 33s 274ms/step - loss: 0.9357 - accuracy: 0.6418\n",
            "Epoch 28/100\n",
            "122/122 [==============================] - 32s 263ms/step - loss: 0.9750 - accuracy: 0.6228\n",
            "Epoch 29/100\n",
            "122/122 [==============================] - 32s 263ms/step - loss: 0.9972 - accuracy: 0.6167\n",
            "Epoch 30/100\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 1.0467 - accuracy: 0.5954\n",
            "Epoch 31/100\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 0.9725 - accuracy: 0.6167\n",
            "Epoch 32/100\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 0.9080 - accuracy: 0.6541\n",
            "Epoch 33/100\n",
            "122/122 [==============================] - 32s 261ms/step - loss: 0.8994 - accuracy: 0.6438\n",
            "Epoch 34/100\n",
            "122/122 [==============================] - 32s 262ms/step - loss: 0.9349 - accuracy: 0.6323\n",
            "Epoch 35/100\n",
            "122/122 [==============================] - 32s 263ms/step - loss: 0.9935 - accuracy: 0.6046\n",
            "Epoch 36/100\n",
            "122/122 [==============================] - 32s 261ms/step - loss: 0.9960 - accuracy: 0.6023\n",
            "Epoch 37/100\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 0.8786 - accuracy: 0.6569\n",
            "Epoch 38/100\n",
            "122/122 [==============================] - 32s 259ms/step - loss: 0.8210 - accuracy: 0.6759\n",
            "Epoch 39/100\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 0.7533 - accuracy: 0.7103\n",
            "Epoch 40/100\n",
            "122/122 [==============================] - 32s 259ms/step - loss: 0.7511 - accuracy: 0.7090\n",
            "Epoch 41/100\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 0.7438 - accuracy: 0.7169\n",
            "Epoch 42/100\n",
            "122/122 [==============================] - 32s 259ms/step - loss: 0.7194 - accuracy: 0.7256\n",
            "Epoch 43/100\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 0.6723 - accuracy: 0.7359\n",
            "Epoch 44/100\n",
            "122/122 [==============================] - 32s 264ms/step - loss: 0.6834 - accuracy: 0.7372\n",
            "Epoch 45/100\n",
            "122/122 [==============================] - 32s 259ms/step - loss: 0.6543 - accuracy: 0.7495\n",
            "Epoch 46/100\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 0.6086 - accuracy: 0.7677\n",
            "Epoch 47/100\n",
            "122/122 [==============================] - 31s 258ms/step - loss: 0.5874 - accuracy: 0.7828\n",
            "Epoch 48/100\n",
            "122/122 [==============================] - 31s 258ms/step - loss: 0.6889 - accuracy: 0.7385\n",
            "Epoch 49/100\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 0.7361 - accuracy: 0.7113\n",
            "Epoch 50/100\n",
            "122/122 [==============================] - 31s 258ms/step - loss: 0.7838 - accuracy: 0.6974\n",
            "Epoch 51/100\n",
            "122/122 [==============================] - 31s 258ms/step - loss: 0.6642 - accuracy: 0.7490\n",
            "Epoch 52/100\n",
            "122/122 [==============================] - 31s 258ms/step - loss: 0.6236 - accuracy: 0.7615\n",
            "Epoch 53/100\n",
            "122/122 [==============================] - 31s 256ms/step - loss: 0.5969 - accuracy: 0.7713\n",
            "Epoch 54/100\n",
            "122/122 [==============================] - 32s 262ms/step - loss: 0.5637 - accuracy: 0.7874\n",
            "Epoch 55/100\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 0.6026 - accuracy: 0.7700\n",
            "Epoch 56/100\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 0.5757 - accuracy: 0.7792\n",
            "Epoch 57/100\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 0.5570 - accuracy: 0.7859\n",
            "Epoch 58/100\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 0.5337 - accuracy: 0.7918\n",
            "Epoch 59/100\n",
            "122/122 [==============================] - 31s 256ms/step - loss: 0.5104 - accuracy: 0.8038\n",
            "Epoch 60/100\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 0.5099 - accuracy: 0.8026\n",
            "Epoch 61/100\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 0.5517 - accuracy: 0.7900\n",
            "Epoch 62/100\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 0.4785 - accuracy: 0.8213\n",
            "Epoch 63/100\n",
            "122/122 [==============================] - 31s 256ms/step - loss: 0.5120 - accuracy: 0.8028\n",
            "Epoch 64/100\n",
            "122/122 [==============================] - 32s 262ms/step - loss: 0.4725 - accuracy: 0.8251\n",
            "Epoch 65/100\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 0.5401 - accuracy: 0.7946\n",
            "Epoch 66/100\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 0.7755 - accuracy: 0.7033\n",
            "Epoch 67/100\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 0.5018 - accuracy: 0.8010\n",
            "Epoch 68/100\n",
            "122/122 [==============================] - 31s 258ms/step - loss: 0.5469 - accuracy: 0.7956\n",
            "Epoch 69/100\n",
            "122/122 [==============================] - 31s 256ms/step - loss: 0.5715 - accuracy: 0.7841\n",
            "Epoch 70/100\n",
            "122/122 [==============================] - 31s 258ms/step - loss: 0.5448 - accuracy: 0.7921\n",
            "Epoch 71/100\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 0.5135 - accuracy: 0.8059\n",
            "Epoch 72/100\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 0.5011 - accuracy: 0.8103\n",
            "Epoch 73/100\n",
            "122/122 [==============================] - 31s 258ms/step - loss: 0.5120 - accuracy: 0.8038\n",
            "Epoch 74/100\n",
            "122/122 [==============================] - 32s 264ms/step - loss: 0.4975 - accuracy: 0.8100\n",
            "Epoch 75/100\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 0.4735 - accuracy: 0.8177\n",
            "Epoch 76/100\n",
            "122/122 [==============================] - 31s 256ms/step - loss: 0.7286 - accuracy: 0.7192\n",
            "Epoch 77/100\n",
            "122/122 [==============================] - 31s 258ms/step - loss: 0.5396 - accuracy: 0.7982\n",
            "Epoch 78/100\n",
            "122/122 [==============================] - 31s 258ms/step - loss: 0.6668 - accuracy: 0.7459\n",
            "Epoch 79/100\n",
            "122/122 [==============================] - 31s 256ms/step - loss: 0.5630 - accuracy: 0.7851\n",
            "Epoch 80/100\n",
            "122/122 [==============================] - 31s 256ms/step - loss: 0.4916 - accuracy: 0.8118\n",
            "Epoch 81/100\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 0.5983 - accuracy: 0.7805\n",
            "Epoch 82/100\n",
            "122/122 [==============================] - 31s 257ms/step - loss: 0.7542 - accuracy: 0.7103\n",
            "Epoch 83/100\n",
            "122/122 [==============================] - 32s 261ms/step - loss: 0.8269 - accuracy: 0.6856\n",
            "Epoch 84/100\n",
            "122/122 [==============================] - 32s 259ms/step - loss: 0.7338 - accuracy: 0.7133\n",
            "Epoch 85/100\n",
            "122/122 [==============================] - 31s 256ms/step - loss: 0.5796 - accuracy: 0.7797\n",
            "Epoch 86/100\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 0.4846 - accuracy: 0.8121\n",
            "Epoch 87/100\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 0.4702 - accuracy: 0.8182\n",
            "Epoch 88/100\n",
            "122/122 [==============================] - 31s 254ms/step - loss: 0.4452 - accuracy: 0.8308\n",
            "Epoch 89/100\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 0.4463 - accuracy: 0.8382\n",
            "Epoch 90/100\n",
            "122/122 [==============================] - 31s 253ms/step - loss: 0.4492 - accuracy: 0.8341\n",
            "Epoch 91/100\n",
            "122/122 [==============================] - 31s 254ms/step - loss: 0.5548 - accuracy: 0.7908\n",
            "Epoch 92/100\n",
            "122/122 [==============================] - 31s 254ms/step - loss: 0.5478 - accuracy: 0.7928\n",
            "Epoch 93/100\n",
            "122/122 [==============================] - 32s 260ms/step - loss: 0.6196 - accuracy: 0.7623\n",
            "Epoch 94/100\n",
            "122/122 [==============================] - 31s 254ms/step - loss: 0.5530 - accuracy: 0.7838\n",
            "Epoch 95/100\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 0.6942 - accuracy: 0.7326\n",
            "Epoch 96/100\n",
            "122/122 [==============================] - 31s 255ms/step - loss: 0.7602 - accuracy: 0.7118\n",
            "Epoch 97/100\n",
            "122/122 [==============================] - 31s 254ms/step - loss: 0.7696 - accuracy: 0.7115\n",
            "Epoch 98/100\n",
            "122/122 [==============================] - 31s 253ms/step - loss: 0.6128 - accuracy: 0.7685\n",
            "Epoch 99/100\n",
            "122/122 [==============================] - 31s 253ms/step - loss: 0.5817 - accuracy: 0.7754\n",
            "Epoch 100/100\n",
            "122/122 [==============================] - 31s 252ms/step - loss: 0.6069 - accuracy: 0.7777\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6RwZu6vQQHsK",
        "colab_type": "code",
        "cellView": "both",
        "colab": {}
      },
      "source": [
        "#Guardar modelo\n",
        "#@markdown optimizer_epochs_samples_features_accuracy_version\n",
        "\n",
        "nombre = \"RNN_SGD_100e_650m_2f_0.77_V1\" #@param {type:\"string\"}\n",
        "model.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.h5')\n",
        "np.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.npy', le.classes_)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "qewdygb-8kK9",
        "colab": {}
      },
      "source": [
        "# Cargar modelo\n",
        "model = tf.keras.models.load_model('Models/model_RNN_SGD_100e_650m_2f_0.77_V1.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "kT8VAHFD8kLG",
        "cellView": "form",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "65f978ea-4e48-4f5a-ea33-42b948d9b5f9"
      },
      "source": [
        "#@title Predicción\n",
        "X_test = x_test.reshape(x_test.shape[0],x_test.shape[1],1)\n",
        "y_predict = np.argmax(model.predict(X_test), axis=1)\n",
        "y_test_single = np.argmax(y_test, axis=1)\n",
        "conf_matrix = confusion_matrix(y_test_single, y_predict)\n",
        "print(conf_matrix)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[ 35   5   9   7  81   4]\n",
            " [  8  76  25  25  51   3]\n",
            " [ 10  28  45  10  72   3]\n",
            " [  2  18  10  57  63  12]\n",
            " [ 58  28  50  26 302  24]\n",
            " [  8   4   1   8  32 100]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "DzyTEaSY8kLJ",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "79f27ac3-cc96-4de5-d546-6b671d62efb8"
      },
      "source": [
        "text = [\"ON\", \"OFF\", \"YES\", \"NO\", \"STOP\", \"UNKNOWN\"]\n",
        "#Vector para gráficas\n",
        "RNN100=[]\n",
        "for i in range(len(text)):\n",
        "  acierto_promedio = conf_matrix[i][i]/(np.sum(conf_matrix[i], axis=0))\n",
        "  RNN100.append(np.round(acierto_promedio, 2))\n",
        "  print(f\"Aciertos clase {text[i]}: {np.round(acierto_promedio*100, 2)}%\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Aciertos clase ON: 24.82%\n",
            "Aciertos clase OFF: 40.43%\n",
            "Aciertos clase YES: 26.79%\n",
            "Aciertos clase NO: 35.19%\n",
            "Aciertos clase STOP: 61.89%\n",
            "Aciertos clase UNKNOWN: 65.36%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BW6BS452InWa",
        "colab_type": "text"
      },
      "source": [
        "### Resultados con RNN\n",
        "Se visualizan los resultados de la predicción de audios (test split) en los modelos entrenados con 50 y 100 Epochs."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2dWP4CqJZJIU",
        "colab_type": "code",
        "cellView": "both",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "def706ed-c429-42f8-bb07-e9181fc84813"
      },
      "source": [
        "# # RNN50=[68.67,52.86,58.33,69.43,82.31]\n",
        "# # RNN100 = [67.63,51.14,61.54,67.52,74.15]\n",
        "\n",
        "# txt=[\"ON\", \"OFF\", \"YES\", \"NO\", \"STOP\", \"UNKNOWN\"]\n",
        "\n",
        "barWidth=0.2\n",
        "plt.bar(np.arange(len(RNN50)), RNN50, color=\"#6A5ACD\", width=barWidth, label=\"50 Epochs\");\n",
        "plt.bar([x + barWidth for x in np.arange(len(RNN50))], RNN100, color=\"#00BFFF\", width=barWidth, label=\"100 Epochs\");\n",
        "plt.xlabel(\"Clases\");\n",
        "plt.ylabel(\"Aciertos en %\");\n",
        "plt.xticks([r + barWidth for r in range(len(text))], txt);\n",
        "plt.title(\"Resultados de la implementación con modelo RNN\");\n",
        "plt.legend();\n",
        "plt.show();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8UdgIhHUPzIW",
        "colab_type": "text"
      },
      "source": [
        "##Entrenamiento Modelo LSTM"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9So-P_Mnal5w",
        "colab_type": "code",
        "cellView": "both",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 353
        },
        "outputId": "900411a3-c1b9-4d8e-b605-88ff2c639941"
      },
      "source": [
        "tf.keras.backend.clear_session()\n",
        "\n",
        "model = tf.keras.models.Sequential(name=\"Modelo_LSTM\")\n",
        "# model.add(tf.keras.layers.Embedding(4, 300, input_length=40))\n",
        "model.add(tf.keras.layers.Input((92,1)))\n",
        "model.add(tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(units=512,\n",
        "                                                          return_sequences=True,\n",
        "                                                          activation='tanh',\n",
        "                                                          name='bilstm_layer1')))\n",
        "model.add(tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(units=512,\n",
        "                                                          return_sequences=True,\n",
        "                                                          activation='tanh',\n",
        "                                                          name='bilstm_layer2')))\n",
        "model.add(tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(units=256,\n",
        "                                                          return_sequences=False,\n",
        "                                                          activation='tanh',\n",
        "                                                          name='bilstm_layer3')))\n",
        "model.add(tf.keras.layers.Dense(units=512, \n",
        "                                   activation='relu',\n",
        "                                   name='dense1') )\n",
        "model.add(tf.keras.layers.Dense(units=256, \n",
        "                                   activation='relu',\n",
        "                                   name='dense2') )\n",
        "model.add(tf.keras.layers.Dense(units = num_labels,\n",
        "                                  activation='softmax',\n",
        "                                  name='ouput_layer'))\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"Modelo_LSTM\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "bidirectional (Bidirectional (None, 92, 1024)          2105344   \n",
            "_________________________________________________________________\n",
            "bidirectional_1 (Bidirection (None, 92, 1024)          6295552   \n",
            "_________________________________________________________________\n",
            "bidirectional_2 (Bidirection (None, 512)               2623488   \n",
            "_________________________________________________________________\n",
            "dense1 (Dense)               (None, 512)               262656    \n",
            "_________________________________________________________________\n",
            "dense2 (Dense)               (None, 256)               131328    \n",
            "_________________________________________________________________\n",
            "ouput_layer (Dense)          (None, 6)                 1542      \n",
            "=================================================================\n",
            "Total params: 11,419,910\n",
            "Trainable params: 11,419,910\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wK9O3CUZcJVy",
        "colab_type": "text"
      },
      "source": [
        "### 50 Epochs"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "cellView": "both",
        "id": "Sk6pSmANhwDh",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "1bd58a61-4fbd-41ce-ccdb-762259678bfb"
      },
      "source": [
        "#@title Entrenamiento\n",
        "epochs = 50 #@param {type:\"slider\", min:50, max:2000, step:50}\n",
        "\n",
        "optimizer = tf.keras.optimizers.SGD(lr=1e-2, momentum=0.9)\n",
        "# optimizer = tf.keras.optimizers.Adam(lr=1e-4)\n",
        "# optimizer = tf.keras.optimizers.Adadelta(lr=1e-4)\n",
        "\n",
        "# optimizer = 'adam' #@param [\"adam\", \"SGD\", \"adadelta\"]\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              metrics=['accuracy'],\n",
        "              optimizer=optimizer)\n",
        "X = x_train.reshape(x_train.shape[0],x_train.shape[1],1)\n",
        "# print(X.shape, y_train.shape)\n",
        "history = model.fit(X, y_train,epochs=epochs,verbose=1)\n",
        "\n",
        "acc = history.history['accuracy']\n",
        "loss = history.history['loss']\n",
        "\n",
        "epochs=range(len(loss)) # Get number of epochs\n",
        "\n",
        "#------------------------------------------------\n",
        "# Plot acc and Loss\n",
        "#------------------------------------------------\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.plot(epochs, acc, 'red', label='acc')\n",
        "plt.title('Acc')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.show();\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.title('Loss')\n",
        "plt.plot(epochs, loss, 'black', label='loss')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Error\")\n",
        "plt.show();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.6468 - accuracy: 0.3713\n",
            "Epoch 2/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.5851 - accuracy: 0.3841\n",
            "Epoch 3/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.5280 - accuracy: 0.4105\n",
            "Epoch 4/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.4904 - accuracy: 0.4241\n",
            "Epoch 5/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.4550 - accuracy: 0.4390\n",
            "Epoch 6/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.4315 - accuracy: 0.4436\n",
            "Epoch 7/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.4204 - accuracy: 0.4572\n",
            "Epoch 8/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.3956 - accuracy: 0.4590\n",
            "Epoch 9/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.3943 - accuracy: 0.4574\n",
            "Epoch 10/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.3708 - accuracy: 0.4615\n",
            "Epoch 11/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.3493 - accuracy: 0.4779\n",
            "Epoch 12/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.3321 - accuracy: 0.4826\n",
            "Epoch 13/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.2887 - accuracy: 0.5008\n",
            "Epoch 14/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.2875 - accuracy: 0.4962\n",
            "Epoch 15/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.2891 - accuracy: 0.4969\n",
            "Epoch 16/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.2635 - accuracy: 0.5056\n",
            "Epoch 17/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.2461 - accuracy: 0.5182\n",
            "Epoch 18/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.2516 - accuracy: 0.5095\n",
            "Epoch 19/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.2323 - accuracy: 0.5226\n",
            "Epoch 20/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.1907 - accuracy: 0.5338\n",
            "Epoch 21/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.1863 - accuracy: 0.5374\n",
            "Epoch 22/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.1767 - accuracy: 0.5433\n",
            "Epoch 23/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.1503 - accuracy: 0.5479\n",
            "Epoch 24/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.1352 - accuracy: 0.5505\n",
            "Epoch 25/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.1257 - accuracy: 0.5577\n",
            "Epoch 26/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.1032 - accuracy: 0.5690\n",
            "Epoch 27/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.0807 - accuracy: 0.5764\n",
            "Epoch 28/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.0692 - accuracy: 0.5864\n",
            "Epoch 29/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.0419 - accuracy: 0.5944\n",
            "Epoch 30/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.0320 - accuracy: 0.5951\n",
            "Epoch 31/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.0102 - accuracy: 0.5982\n",
            "Epoch 32/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.9758 - accuracy: 0.6164\n",
            "Epoch 33/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.9371 - accuracy: 0.6274\n",
            "Epoch 34/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.9073 - accuracy: 0.6521\n",
            "Epoch 35/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.8851 - accuracy: 0.6456\n",
            "Epoch 36/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.8873 - accuracy: 0.6495\n",
            "Epoch 37/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.8371 - accuracy: 0.6685\n",
            "Epoch 38/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.8158 - accuracy: 0.6856\n",
            "Epoch 39/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.7595 - accuracy: 0.7041\n",
            "Epoch 40/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.7381 - accuracy: 0.7141\n",
            "Epoch 41/50\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.6944 - accuracy: 0.7254\n",
            "Epoch 42/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.6680 - accuracy: 0.7323\n",
            "Epoch 43/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.6260 - accuracy: 0.7579\n",
            "Epoch 44/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.5601 - accuracy: 0.7818\n",
            "Epoch 45/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.5041 - accuracy: 0.8033\n",
            "Epoch 46/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.5105 - accuracy: 0.7987\n",
            "Epoch 47/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.4562 - accuracy: 0.8259\n",
            "Epoch 48/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.5000 - accuracy: 0.8128\n",
            "Epoch 49/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.3712 - accuracy: 0.8644\n",
            "Epoch 50/50\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.3639 - accuracy: 0.8687\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfEAAAGDCAYAAAA72Cm3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deZyd4/3/8ddHJPZ9aS0hqaUaLRGp8kVKKKqE2kOVVu1bFbWVatSSLtQSVW0ttTQSW0NTRMSPKpIJopIIkSKxRKIkVPZcvz+uExmR5SSZM/c5c17Px+M8Zu773DPzmZvxdl33tURKCUmSVHuWKboASZK0ZAxxSZJqlCEuSVKNMsQlSapRhrgkSTXKEJckqUYZ4pIk1ShDXNKnIuLxiPggIpYruhZJi2aISwIgItoBOwMJ6FZoMZLKYohLmuP7wDPALcBRc05GRNuIuDciJkTE+xFxXaP3jo2IkRHxUUSMiIhOzV+2VL+WLboASVXj+8CVwLPAMxHxBWAi8CDwGHAkMAvoDBARBwMXA/sDDcAmwIxmr1qqY+Ha6ZIiYidgELBeSmliRLwM/IHcMu9XOj9znq95GOifUrq62QuWBNgSl5QdBTySUppYOr6zdO4t4I15A7ykLfBaM9UnaT4McanORcQKwCFAq4h4t3R6OWB1YDywUUQsO58gH0vuQpdUEAe2Sdqf/Ky7A9Cx9PoK8GTpvXeAKyJipYhYPiJ2LH3dn4CzImLbyDaNiI0LqF+qW4a4pKOAm1NKb6aU3p3zAq4DugP7ApsCbwLjgEMBUkp9gUvJXe8fAfcDaxZQv1S3HNgmSVKNsiUuSVKNMsQlSapRhrgkSTXKEJckqUYZ4pIk1aiaW+xl7bXXTu3atSu6DEmSmsXQoUMnppTWmd97NRfi7dq1o6GhoegyJElqFhHxxoLesztdkqQaZYhLklSjDHFJkmqUIS5JUo0yxCVJqlGGuCRJNcoQlySpRhnikiTVKENckqQaZYhLklSjDHFJkmqUIS5JUlMZOxb692+2H1dzG6BIklQ1UoIRI+C+++D++2HoUFhpJXj/fVhuuYr/eENckqTFMXs2PPPM3OAePTqf3357uOIK2H//ZglwMMQlSSrfNdfAZZfB+PHQujV07QpnngndusH66zd7OYa4JEnlmDoVzjkHttoKrroK9t4bVlut0JIMcUmSyvHPf+Ygv+gi+M53iq4GcHS6JEnlGTAgd6F/85tFV/IpQ1ySpHIMGAA77AArr1x0JZ8yxCVJWpQJE+D55+Fb3yq6ks8wxCVJWpSBA/NHQ1ySpBozYACsvjp07lx0JZ9hiEuStDAp5RDv2hVatSq6ms8wxCVJWphXXslroldZVzoY4pIkLdyAAfmjIS5JUo0ZMADat4dNNim6ks8xxCVJWpAZM2DQoKpshYMhLknSgg0eDB99ZIhLklRzBgyAiDwyvQoZ4pIkLciAAfD1r8OaaxZdyXwZ4pIkzc+kSfDss1XblQ6GuCRJ8/f44zBrliEuSVLNGTAAVlop71xWpQxxSZLmZ8CAvHd4mzZFV7JAhrgkSfN688283GoVd6WDIS5J0udV8VKrjRnikiTNa8AAWH996NCh6EoWyhCXJNWH55+H738f3n134dfNng0DB8Luu+eFXqqYIS5JavlSglNPhdtug513hjfeWPC1L7wAEydWfVc6GOKSpHrwyCPw1FNwwgk5oHfaCUaNmv+1c56H775789W3hAxxSVLLlhJceCG0awdXX50XcZk+PbfIX3jh89cPGABf+xp88YvNXeliM8QlSS3bgw/CkCE5yNu0ga23hiefhOWXh112gX/9a+61U6bAP/9ZE13pYIhLklqy2bPhootgk03gyCPnnt988xzW666bA3tOF/qTT8K0aTUT4ssWXYAkSRVz3325y/wvf4HWrT/73kYb5dDeYw/YZx/o3Tu3ytu0gS5diql3MUVKqegaFkvnzp1TQ0ND0WVIkqrdrFm563zWLHjpJWjVav7XffAB7L03DB4Ma6wBW20Fjz3WvLUuREQMTSl1nt97dqdLklqmPn1g+HC4+OIFBzjk4B4wID8ff//9mulKB1vikqSWaOZM2HJLWG653J2+TBlt1qlT4ZZb4PDDYdVVK15iuRbWEveZuCSp5bnjjryByb33lhfgkEern3BCZetqYnanS5JalhkzoEcP6NQJ9t+/6Goqypa4JKllufVWGDMmzw+v8rXPl1ZFW+IRsVdEjIqI0RFx7nze3ygiBkXE8xHxYkTsXcl6JEk17N//zvt8L8y0aXDJJfCNb+QR5y1cxUI8IloBvYBvAx2A7hEx755uPwP6pJS2AQ4Drq9UPZKkGvbii7l7fOONYfvt4be/nf8mJn/+cw76Hj1afCscKtsS3w4YnVIak1KaDvQG9pvnmgTMGQK4GvB2BeuRJNWimTPhmGPyVLBf/jKve37WWXkt9MaBPmUKXHpp3tykhqaJLY1KhvgGwNhGx+NK5xq7GPheRIwD+gOnzu8bRcRxEdEQEQ0TJkyoRK2SpGr1u99BQwNcey1ccAE89xy8+ipcfnkexDYn0Dt0gLffzt3pddAKh+JHp3cHbkkpbQjsDdwWEZ+rKaV0Y0qpc0qp8zrrrNPsRUqSCvLaa3nt8333hUMOmXt+003h3HNh6FAYPRquuALWXjvP8d5ll8LKbW6VHJ3+FtC20fGGpXONHQPsBZBSejoilgfWBt6rYF2SpFqQEhx3XF7z/PrrF9y63mQTOOec/KozlWyJDwE2i4j2EdGGPHCt3zzXvAnsBhARXwGWB+wvlyTBTTflNcx/9SvYcMOiq6lKFQvxlNJM4BTgYWAkeRT68IjoERHdSpedCRwbEcOAvwJHp1pbB1aS1PTeeQfOPDPvJnbssUVXU7UquthLSqk/ecBa43MXNfp8BLBjJWuQJNWgU07Ja5n/8Y/lL5tah1yxTZJUXe65J695fvnlsPnmRVdT1fzfG0lS9fjgg9wK79gxd6droWyJS5Kqx1lnwYQJ8Pe/51HpWihb4pKk6jBwYB6RfuaZeYlVLZIhLkkq3ief5Dnhm24KF19cdDU1w+50SVLxrrgibx86aBCssELR1dQMW+KSpGL95z95QZfu3etqydSmYIhLkop11lnQqlUOci0WQ1ySVJzHHstzws87z6VVl4AhLkkqxsyZcPrpeRtR54QvEQe2SZKKccMN8NJLuSXuYLYlYktcktT8Jk7M+4Tvthvsv3/R1dQsQ1yS1PwuvBAmT4arr17wPuFaJENcktS8hg2DG2+Ek06CLbcsupqaZohLkppPSnDaabDGGvCLXxRdTc1zYJskqfn07QtPPJEHta2xRtHV1Dxb4pKk5vHJJ3lhl44d4Uc/KrqaFsGWuCSpefTsCWPHwh135BXatNRsiUuSKu+NN/KyqoceCjvvXHQ1LYYhLkmqvHPPzVPJfv3roitpUQxxSVJlNTRA797wk59A27ZFV9OiGOKSpMpJCc45B9ZaC37606KraXEc2CZJqpxHHsk7lf3ud7DqqkVX0+LYEpckVcbs2bkV3r49nHBC0dW0SLbEJUmVceedeYnVO++E5ZYrupoWyZa4JKnpTZ0KP/sZdOqUp5WpImyJS5Ka3vXX57nhf/oTLGN7sVK8s5KkpvXhh3DppbDHHrD77kVX06IZ4pKkptWzJ3zwQf6oijLEJUlNZ9y4PJ3siCPyRieqKENcktR0Lr44Ty275JKiK6kLhrgkqWmMGAE33wwnnwzt2hVdTV0wxCVJTePcc2GVVeCCC4qupG4Y4pKkpffkk/DAAznI11qr6GrqhiEuSVpys2blFdmOPBI22ABOO63oiuqKIS5JWnyzZsEdd8CWW+aR6KuskrcbXXHFoiurK4a4JKl8M2fC7bdDhw7wve9BmzZw9915jfSddiq6urpjiEuSFm3mTLjtthzeRx4Jyy8P99wDL7wABx7o0qoFce10SdLCDRuWNzEZNQq23hruvRf228/grgKGuCRpwfr2haOPhtVXzy3v/fc3vKuI/yQkSZ83ezZceCEcckhufTc0wAEHGOBVxpa4JOmzJk/Og9YeeACOOQZ69YLlliu6Ks2HIS5JmuvVV/Pz7ldegWuvzUuoRhRdlRbAEJckZQ8/DIcdBq1awYABsOuuRVekRfDhhiTVu5TgN7+BvfeGjTaCIUMM8BphiEtSPfvvf+Ggg+Dss/PAtX/9C9q3L7oqlckQl6R69fjjsNVWeQDbr38NffrASisVXZUWgyEuSfVmxoy8XWjXrjm0n3kGzjrLAWw1yIFtklRPxoyBww+HZ5/N08d+9ztYeeWiq9ISMsQlqV7cfjucdFIefd6nDxx8cNEVaSnZnS5JLd2cxVuOPBI6dsxroRvgLUJFQzwi9oqIURExOiLOnc/7V0XEC6XXKxHxYSXrkaS6M2MGfPvbea/vHj1g0KA8jUwtQsW60yOiFdAL+BYwDhgSEf1SSiPmXJNSOqPR9acC21SqHkmqS+edl6eN9e6ddyJTi1LJlvh2wOiU0piU0nSgN7DfQq7vDvy1gvVIUn3p1w9++9v8HNwAb5EqGeIbAGMbHY8rnfuciNgYaA88toD3j4uIhohomDBhQpMXKkktzhtv5C1EO3WCK68suhpVSLUMbDsMuDulNGt+b6aUbkwpdU4pdV5nnXWauTRJqjHTp+ctRGfNyqPQ3YGsxarkFLO3gLaNjjcsnZufw4CTK1iLJNWPc8+FwYPh7rthk02KrkYVVMmW+BBgs4hoHxFtyEHdb96LImILYA3g6QrWIkn14f774aqr4NRT4cADi65GFVaxEE8pzQROAR4GRgJ9UkrDI6JHRHRrdOlhQO+UUqpULZJUF/7zn/wcvHPnvBa6WryKrtiWUuoP9J/n3EXzHF9cyRokqS5Mnz53BLrPweuGy65KUktw9tl5H/B773Ur0TpSLaPTJUlL6p574Jpr4Mc/hu9+t+hq1IwMcUmqZaNGwQ9/CNttBz17Fl2NmpkhLkm1atIk2G+//Py7b19o06boitTMfCYuSbVo1iw44gh47TUYONBNTeqUIS5Jteiii+Dvf4devaBLl6KrUUHsTpekWtO3L1x2GfzoR3DiiUVXowIZ4pJUS4YNywu67LADXHcdRBRdkQpkiEtSrZg4MQ9kW331PK3MBV3qns/EJakWzJiRdyZ791144glYb72iK1IVMMQlqRacdRYMGgS33prnhEvYnS5J1e/mm+euyPb97xddjaqIIS5J1WzoUDjhBNhtN3cm0+cY4pJUrT76CA47DNZdF3r3hmV9AqrP8t8ISapGKeU54GPGwOOPw9prF12RqpAtcUmqRn/5C9xxB/z857DzzkVXoypliEtStXnlFTj5ZPjmN+GCC4quRlXMEJekajJtGhx6KCy/PNx+O7RqVXRFqmI+E5ekavLTn8ILL0C/frDhhkVXoypnS1ySqkW/fnk++Omnw777Fl2NaoAhLknVYNw4+MEPYJttoGfPoqtRjVhkiEfEvhFh2EvS/EyenKeDLY1Zs+CII/Lz8N693dhEZSsnnA8FXo2IX0XEFpUuSJJqxiuvwBe/CFttlaeETZ++ZN/nl7/Mm5pcfz1svnnT1qgWbZEhnlL6HrAN8BpwS0Q8HRHHRcQqFa9OkqrZ2WfnVdQi4KijYJNN4Mor80pri5ISjBwJv/oV9OgBRx7puuhabGV1k6eUJgN3A72B9YDvAs9FxKkVrE2Sqtdjj+WBaOefD8OGQf/+sOmmcOaZ0LZtPv/uu5/9mg8+gLvvhmOPhY03hg4d4Jxz4BvfgF69ivk9VNMiLeJZTkR0A34AbAr8Bbg1pfReRKwIjEgptat4lY107tw5NTQ0NOePlKTPmjULtt0WPvwQXn45z+meY/DgvFHJPfdA69a5hb7BBvDww/DsszB7Nqy6at7QZM89YY89oH374n4XVb2IGJpS6jy/98qZJ34gcFVK6YnGJ1NKn0TEMU1RoCTVlFtvza3v3r0/G+CQ9/ru2xdefTV3rd98c35W/vWv59b5nnvmlnfr1sXUrhalnJZ4e+CdlNLU0vEKwBdSSq9XvrzPsyUuqVAffZQHn7VvD089lZ+HL8yHH+bW95prNk99anEW1hIv55l4X2B2o+NZpXOSVH9+9av8rPuqqxYd4ACrr26Aq2LKCfFlU0qfzpsofd6mciVJUpV68034zW/g8MNzl7hUsHJCfEJpcBsAEbEfMLFyJUlSlTr//Pzx8suLrUMqKWdg2wnAHRFxHRDAWMDJjJLqy+DBeX/vCy6AjTYquhoJKCPEU0qvAdtHxMql448rXpUkVZOU4Iwz8ups55xTdDXSp8raijQivgNsCSwfpYEcKaUeFaxLkqpH377wr3/Bn/4Eq7hYpapHORug3EBeP/1Ucnf6wcDGFa5LkqrD1Km59b311nD00UVXI31GOS3x/0spbRURL6aUfhERvwX+UenCJKkqXH01vP46PPootGpVdDXSZ5QzOn1q6eMnEbE+MIO8froktWxvvw2XXgr77puXSZWqTDkh/kBErA78GngOeB24s5JFSVLhhgyB7beHmTPzWuhSFVpoiEfEMsDAlNKHKaV7yM/Ct0gpXdQs1UlSEW66CXbeGZZZBv75T/jyl4uuSJqvhYZ4Smk20KvR8bSU0qSKVyVJRZg+HU46CY45Jod4QwN06lR0VdICldOdPjAiDowoZ5FgSapRb78Nu+4Kv/89/PSn8I9/wNprF12VtFDljE4/HvgJMDMippKnmaWU0qoVrUySmstTT8FBB+Udyu66Cw45pOiKpLIssiWeUlolpbRMSqlNSmnV0rEBLqn2pQTXXw+77AIrrQTPPGOAq6YssiUeEV3mdz6l9ETTlyNJzejcc/PWonvvnddFX331oiuSFks53elnN/p8eWA7YCjQtSIVSVJz6Ns3B/jxx+fW+DLlDBGSqks5G6Ds2/g4ItoCv6tYRZJUaa+8kkegb789XHONAa6atST/5o4DvtLUhUhSs/jkEzj4YGjdOg9ia9Om6IqkJVbOM/FrgVQ6XAboSF65TZJqzymnwL//Df37uy+4al45z8QbGn0+E/hrSumpCtUjSZVz001w883ws5/BXnsVXY201MoJ8buBqSmlWQAR0SoiVkwpfVLZ0iSpCQ0bBiefDF27wsUXF12N1CTKWrENWKHR8QrAo+V884jYKyJGRcToiDh3AdccEhEjImJ4RLixiqSmN2lSXsxljTXgzjvdUlQtRjkt8eVTSh/POUgpfRwRKy7qiyKiFXnd9W+RB8MNiYh+KaURja7ZDDgP2DGl9EFErLvYv4EkLUxKeST6f/4DgwbBF75QdEVSkymnJf6/iPh0B4CI2BaYUsbXbQeMTimNSSlNB3oD+81zzbFAr5TSBwAppffKK1uSynTNNXDPPXD55XlTE6kFKacl/mOgb0S8TV43/YvAoWV83QbA2EbH44BvzHPN5gAR8RTQCrg4pfRQGd9bkhbt6afhrLOgW7f8UWphylnsZUhEbAHM2VB3VEppRhP+/M2AXYANgSci4msppQ8bXxQRxwHHAWzklBBJC/L22zm4n3kmf2xogLZt4ZZbwI0Y1QKVM0/8ZOCOlNJLpeM1IqJ7Sun6RXzpW0DbRscbls41Ng54tvQ/Bf+JiFfIoT6k8UUppRuBGwE6d+6ckFSfUoIpU2Dy5PyaMAGGDJkb3G++ma9r0wa23TbvDX7SSXlAm9QCldOdfmxKqdecg9IAtGOBRYX4EGCziGhPDu/DgMPnueZ+oDtwc0SsTe5eH1Nu8ZJaqBkz4IYb8vrmH344N7QnT4ZZsz5//UYbwQ47wBln5I8dO8JyyzV/3VIzKyfEW0VEpJQSfDrqfJHrFKaUZkbEKcDD5OfdN6WUhkdED6AhpdSv9N4eETECmAWcnVJ6f0l/GUktwD/+AT/5Cbz8MnTqBJtsAquuOve12mpzP1999RzY669fdNVSIcoJ8YeAuyLiD6Xj44F/lPPNU0r9gf7znLuo0ecJ+EnpJamejRwJZ56ZQ3yzzaBfP9hnH59lSwtRToifQx5UdkLp+EXyCHVJWnr//S/84hfQqxesvDL89rd5fXM3JpEWaZHzxFNKs4FngdfJc7+7AiMrW5akFm/GDLjuutzqvu46OPZYePXV3JVugEtlWWBLPCI2Jw866w5MBO4CSCnt2jylSWqxRo6Eww+HF17Ia5lfdRVstVXRVUk1Z2Hd6S8DTwL7pJRGA0TEGc1SlaSWKSX44x/hxz+GlVbKK6l997s+95aW0MK60w8A3gEGRcQfI2I38optkrT43n8fDjwQjj8edtoJXnwRDjjAAJeWwgJDPKV0f0rpMGALYBB5+dV1I+L3EbFHcxUoqQUYNAi23hoefBB+8xt46CFYb72iq5JqXjkD2/6XUrozpbQvedW158kj1iVp4WbMgAsugN12y93nzzyTp5EtU87eS5IWZbH+klJKH6SUbkwp7VapgiS1EK+9lrvNL7sMfvhDGDo0L94iqcmUM09ckso3eXLuMr/ySmjdGvr0gYMPLroqqUWyT0tS05g2LU8V+9KX4JJL8mprw4YZ4FIFGeKSls6sWXDrrbD55nmhlk6d8hagvXvnjUkkVYwhLmnJpAQPPJA3IDn6aFh3XXj0UXjkkbwNqKSKM8QlLb5Ro6BLF+jWLXej9+kDgwfnUeiSmo0D2yQtnv79oXv3vL75DTfkkeetWxddlVSXbIlLKk9K0LNnHrC2ySZ5ytjxxxvgUoFsiUtatE8+gWOOyYPVDjsM/vxnWHHFoquS6p4tcUkL9+abedGWu+6CK66AO+80wKUqYUtc0oI98QQcdBBMn57XPd9776IrktSILXFJ83fDDXm0+ZprwrPPGuBSFTLEJX3Wyy/nVdZOPBH22CMH+Je/XHRVkubDEJeUjRmTF23Zcsu8Vegll0C/frDaakVXJmkBfCYu1btx4+CXv8wjzpddFs44A845B9ZZp+jKJC2CIS7Vq/Hj82jz3/8eZs/Oc77PPx/WX7/oyiSVyRCX6sW0aTB8eN5ZbMiQvGnJtGm5C/3CC2HjjYuuUNJiMsSllmjCBHjhhfwaNiy/Xn4ZZs7M76+wAnz3u/Dzn+fdxyTVJENcainGjYO774a+feFf/5p7foMN8k5j3brB1lvn16abQqtWxdUqqUkY4lItGzt2bnA//XQ+t/XWeWT5//1f/nyttYqtUVLFGOJSrRk/Pi992qcPPPNMPtexI1x6aZ7fvdlmxdYnqdkY4lKtePZZuPbaHN4zZsA228Bll+VlUQ1uqS4Z4lI1mzYtbzxy3XV5RPkqq+SV1E48EbbYoujqJBXMEJeq0bhxee3yG2/MI82/8hXo1QuOPDIHuSRhiEvVYdasPA3sscdg4EAYMCAvwNKtG5xySt6IJKLoKiVVGUNcKkJKMHJkDu3HHoPHH4cPPsjvfeUrcOaZucu8Xbsiq5RU5QxxqTlNmgRnnw0PPADvvpvPbbxxXnila1fYdVeXPZVUNkNcai6jRsF++8Frr+WpYLvtloO7ffuiK5NUowxxqTn07w/du8Nyy+Vn3l26FF2RpBbA/cSlSkoJevaEffaBL30pTxMzwCU1EVviUqV88gkccwz07g2HHgo33QQrrlh0VZJaEFviUiW8+SbstFNeqOXyy+GvfzXAJTU5W+JSU3vySTjwwLza2oMPwt57F12RpBbKlrjUlO64I484X2MNGDzYAJdUUYa41FR+//u8LOrOO+fNSr785aIrktTCGeJSY88/D2+8sfhfd8UVcNJJeRR6//6w+upNX5skzcMQl+a4+Wb4+tfz7mCXX563+1yUlOC88/Kre3e45x5YfvnK1ypJGOJSDuIrroAf/jAve7r33nD++dCpEzz99IK/bvbsvDnJFVfAccfBbbdB69bNV7ekumeIq77Nng0//vHclvTf/55b03/7G3z4Iey4Y96I5MMPP/t1M2fCUUfB9dfntdBvuAFatSrmd5BUtwxx1a9p0+CII+Caa3KQ3347tGmT3+vWDUaMgNNPz3t6f+Ur0KdPbrVPnQoHHZSv/+Uv84psbhMqqQCGuOrT5Mnwne/k1dR69oQrr4Rl5vlzWGUVuOqqPFVs/fXzqmv77JNff/tbDv8LLjDAJRXGxV5Uf8aPh29/G158EW69Fb7//YVfv+22ecrYtdfChRfClClwyy25O12SCmSIq7689hrsuSe88w7061f+YizLLgtnnAGHHALvvQfbbFPZOiWpDIa46sd998EJJ+RBaQMHwvbbL/732GCD/JKkKlDRZ+IRsVdEjIqI0RFx7nzePzoiJkTEC6XXjypZj+rU2LGw//5wwAHwhS/AU08tWYBLUpWpWIhHRCugF/BtoAPQPSI6zOfSu1JKHUuvP1WqHtWhWbPg6quhQwd45JE8gG3o0LyYiyS1AJVsiW8HjE4pjUkpTQd6A/tV8Oeplr33Hnzta/DNb8J11+Vn1kvjuefgG9/IU8d22gmGD4ef/tTFWCS1KJUM8Q2AsY2Ox5XOzevAiHgxIu6OiLbz+0YRcVxENEREw4QJEypRq4o0a1aerz16NEycCKeemp87d+mSR4S//Xb53+vjj+HMM/PyqePG5Slk/ftD+/aVq1+SClL0wLYHgL+mlKZFxPHArUDXeS9KKd0I3AjQuXPn1LwlquJ69IBHH4U//QmOOSYvstK3b36ddlpecGXHHeHgg6FjR/joozzPe/JkmDRp7ueTJ8Njj+Vn4Mcfn9c/X2ONon87SaqYSKkymRgROwAXp5T2LB2fB5BSunwB17cC/ptSWm1h37dz586poaGhqctVUR55BPbaK8/Vvvnmzy+cMnLk3EB/6aX5f49lloFVV82vjTfO4b3jjpWvXZKaQUQMTSl1nu97FQzxZYFXgN2At4AhwOEppeGNrlkvpfRO6fPvAueklBY6bNgQb0HGjs3zrddfH555BlZcceHXjxqVv2a11eaG9qqr5q9z1TRJLdTCQrxi3ekppZkRcQrwMNAKuCmlNDwiegANKaV+wGkR0Q2YCfwXOLpS9ajKzJiRlzGdNi23shcV4ABf/nJ+SZKACj8TTyn1B/rPc+6iRp+fB5xXyRpUpc45J2/zedddBrMkLSE3QF8BphEAAA17SURBVFHzu/fevLHIqafmZUwlSUvEEFfzGj0afvAD2G47+M1viq5GkmqaIa7mM2VK3od72WXz3txz9u6WJC2RoueJq56cdhoMG5YXX9l446KrkaSaZ4irsoYPnzvPe8QIOP/8vJe3JGmpGeJqWil9NrhHjsxzuHfeGX7/ezj22KIrlKQWwxBX0xg/Hnr1ysH98ss5uLt0gZNPzluArrde0RVKUotjiGvpTZoEXbvm8O7SJU8dO+AA+OIXi65Mklo0Q1xLZ+bMPNf7lVfyJia77lp0RZJUNwxxLbmU8ojzRx7JO5AZ4JLUrJwnriV37bV5sNrZZ+ctRCVJzcoQ15L5+9/hjDNg//3hiiuKrkaS6pIhrsX34otw2GHQsSPcfnvez1uS1Oz8r68Wz7vvwr775n28+/WDlVYquiJJqlsObFP5pkzJ3ecTJ8KTT8IGGxRdkSTVNUNc5Zk9G44+GgYPzluJdupUdEWSVPfsTtfCpQTPPQfHHZd3HuvZM7fGJUmFsyVeD1KCGTPK3/rz449h4EB48MG849jbb+dlVE89Fc46q7K1SpLKZoi3FP/7H/znPwt+ffQRrLkmbLghtG372Y8bbpjfe+qpHNyDBsH06Xnw2p57wne+k3ceW3fdon9LSVIjhnitmzoVjjgiP6dubKWVoH37/NplF1h7bXjnHRg3Lr+efTYPUJvX5pvDKafk4N5pp/Jb75KkZmeI17JPPsnPpx99NHdzb7stfOlLObjXXjt3gS/MlCnw1ls51MePz4PVNtuseWqXJC01Q7xW/e9/eb7244/DTTflkeOLa4UVYNNN80uSVHMM8Vr00Ue5u/upp+C223J3uiSp7hjitWbSpDzIbPBg+Otf8zagkqS6ZIjXkg8+yKPFn38+z9k+4ICiK5IkFcgQrxXvvw977AEvvZRHou+7b9EVSZIKZojXggkTYPfdYdQouP/+3J0uSap7hng1Gz4c+vaFW27JU8AeeAC+9a2iq5IkVQlDvNrMCe6+fWHEiDzXe+ed8yj0nXcuujpJUhUxxKvBiBFw1105uEeOzMHdpQtcd10evLbeekVXKEmqQoZ40a67Lm8sMie4Tz7Z4JYklcUQL9Jtt+UA79YN/vAH+OIXi65IklRDDPGi/O1v8IMfQNeuuSt9+eWLrkiSVGOWKbqAujRoEBx6aN6w5P77DXBJ0hIxxJvb4MG5+3zTTaF/f1hllaIrkiTVKEO8OQ0fnhdqWWcdeOQRWGutoiuSJNUwQ7y5jBmTF2pZbrm8//f66xddkSSpxjmwrTm8804O8KlT4Ykn4EtfKroiSVILYIhX2n//mzcuGT8eBg6Er3616IokSS2EIV5J77+fW+CvvJIHsX3jG0VXJElqQQzxSmm889h998FuuxVdkSSphTHEK+Hdd3NojxnjzmOSpIoxxJva22/nVdjGjs1d6LvuWnRFkqQWyhBvSmPH5gB/91146CG3DpUkVZQh3lRefz0H+Pvv54Vcdtih6IokSS2cId4UxozJ3eaTJ+eFXL7+9aIrkiTVAUN8ab36ag7wKVPgscdgm22KrkiSVCcM8aUxYwbssw9Mm5Z3Jttqq6IrkiTVEUN8adx8c17IpV8/A1yS1OzcAGVJTZkCv/hFHsC2zz5FVyNJqkMVDfGI2CsiRkXE6Ig4dyHXHRgRKSI6V7KeJtWrV54TfvnlEFF0NZKkOlSxEI+IVkAv4NtAB6B7RHSYz3WrAKcDz1aqliY3eXIO7z33hG9+s+hqJEl1qpIt8e2A0SmlMSml6UBvYL/5XHcJ0BOYWsFamtZvf5t3J7v00qIrkSTVsUqG+AbA2EbH40rnPhURnYC2KaW/V7COpjVhAlx5JRx0EGy7bdHVSJLqWGED2yJiGeBK4Mwyrj0uIhoiomHChAmVL25hLrsMPvkELrmk2DokSXWvkiH+FtC20fGGpXNzrAJ8FXg8Il4Htgf6zW9wW0rpxpRS55RS53XWWaeCJS/Cm2/C9dfDUUfBFlsUV4ckSVQ2xIcAm0VE+4hoAxwG9JvzZkppUkpp7ZRSu5RSO+AZoFtKqaGCNS2dHj3yx4svLrQMSZKggiGeUpoJnAI8DIwE+qSUhkdEj4joVqmfWzGjRuXFXU48ETbaqOhqJEkiUkpF17BYOnfunBoaCmisH3JI3h98zBhYd93m//mSpLoUEUNTSvNdR8UV28rx3HPQty+ccYYBLkmqGoZ4OS64ANZYA846q+hKJEn6lBugLMoTT8BDD0HPnrDaakVXI0nSp2yJL8zs2XD++bDeenDKKUVXI0nSZ9gSX5AZM+CYY+Cpp+DGG2HFFYuuSJKkzzDE52fKFDj0UHjggTw3/Ec/KroiSZI+xxCf16RJsO++8M9/5tXZTjyx6IokSZovQ7yx8eNhr73gpZfgzjvhsMOKrkiSpAUyxOd4/XX41rfg7bdzN/peexVdkSRJC2WIQ25577lnfhb+6KOwww5FVyRJ0iI5xezpp6FLF0gpzwk3wCVJNaK+Q3zAANh9d1hzzTyV7KtfLboiSZLKVt8hvvLKsM02eSR6+/ZFVyNJ0mKp72fiO+wATz4JEUVXIknSYqvvljgY4JKkmmWIS5JUowxxSZJqlCEuSVKNMsQlSapRhrgkSTXKEJckqUYZ4pIk1ShDXJKkGmWIS5JUowxxSZJqlCEuSVKNMsQlSapRkVIquobFEhETgDea8FuuDUxswu9Xz7yXTcd72XS8l03He9k0Fvc+bpxSWmd+b9RciDe1iGhIKXUuuo6WwHvZdLyXTcd72XS8l02jKe+j3emSJNUoQ1ySpBpliMONRRfQgngvm473sul4L5uO97JpNNl9rPtn4pIk1Spb4pIk1ai6DvGI2CsiRkXE6Ig4t+h6aklE3BQR70XES43OrRkRAyLi1dLHNYqssRZERNuIGBQRIyJieEScXjrvvVxMEbF8RAyOiGGle/mL0vn2EfFs6e/8rohoU3SttSIiWkXE8xHxYOnYe7kEIuL1iPh3RLwQEQ2lc03yN163IR4RrYBewLeBDkD3iOhQbFU15RZgr3nOnQsMTCltBgwsHWvhZgJnppQ6ANsDJ5f+PfReLr5pQNeU0tZAR2CviNge6AlclVLaFPgAOKbAGmvN6cDIRsfeyyW3a0qpY6OpZU3yN163IQ5sB4xOKY1JKU0HegP7FVxTzUgpPQH8d57T+wG3lj6/Fdi/WYuqQSmld1JKz5U+/4j8H8wN8F4utpR9XDpsXXoloCtwd+m897JMEbEh8B3gT6XjwHvZlJrkb7yeQ3wDYGyj43Glc1pyX0gpvVP6/F3gC0UWU2sioh2wDfAs3sslUur+fQF4DxgAvAZ8mFKaWbrEv/Py/Q74KTC7dLwW3ssllYBHImJoRBxXOtckf+PLNkV10rxSSikinPpQpohYGbgH+HFKaXJu9GTey/KllGYBHSNideA+YIuCS6pJEbEP8F5KaWhE7FJ0PS3ATimltyJiXWBARLzc+M2l+Ruv55b4W0DbRscbls5pyY2PiPUASh/fK7iemhARrckBfkdK6d7Sae/lUkgpfQgMAnYAVo+IOQ0W/87LsyPQLSJeJz9q7ApcjfdyiaSU3ip9fI/8P5fb0UR/4/Uc4kOAzUqjLdsAhwH9Cq6p1vUDjip9fhTwtwJrqQml54x/BkamlK5s9Jb3cjFFxDqlFjgRsQLwLfIYg0HAQaXLvJdlSCmdl1LaMKXUjvzfxsdSSkfgvVxsEbFSRKwy53NgD+AlmuhvvK4Xe4mIvcnPfVoBN6WULi24pJoREX8FdiHvxjMe+DlwP9AH2Ii809whKaV5B7+pkYjYCXgS+Ddznz2eT34u7r1cDBGxFXmAUCtyA6VPSqlHRHyJ3JpcE3ge+F5KaVpxldaWUnf6WSmlfbyXi690z+4rHS4L3JlSujQi1qIJ/sbrOsQlSapl9dydLklSTTPEJUmqUYa4JEk1yhCXJKlGGeKSJNUoQ1yqAxExq7SD0pxXk22oEhHtGu9mJ6n5uOyqVB+mpJQ6Fl2EpKZlS1yqY6V9jn9V2ut4cERsWjrfLiIei4gXI2JgRGxUOv+FiLivtGf3sIj4v9K3ahURfyzt4/1IacU0IuK00l7pL0ZE74J+TanFMsSl+rDCPN3phzZ6b1JK6WvAdeQVDAGuBW5NKW0F3AFcUzp/DfD/Snt2dwKGl85vBvRKKW0JfAgcWDp/LrBN6fucUKlfTqpXrtgm1YGI+DiltPJ8zr8OdE0pjSltxPJuSmmtiJgIrJdSmlE6/05Kae2ImABs2HipzdIWqgNSSpuVjs8BWqeUfhkRDwEfk5fkvb/Rft+SmoAtcUlpAZ8vjsbrZ89i7nib7wC9yK32IY12wJLUBAxxSYc2+vh06fN/kXevAjiCvEkLwEDgRICIaBURqy3om0bEMkDblNIg4BxgNeBzvQGSlpz/VyzVhxUi4oVGxw+llOZMM1sjIl4kt6a7l86dCtwcEWcDE4AflM6fDtwYEceQW9wnAu8s4Ge2Am4vBX0A15T2+ZbURHwmLtWx0jPxzimliUXXImnx2Z0uSVKNsiUuSVKNsiUuSVKNMsQlSapRhrgkSTXKEJckqUYZ4pIk1ShDXJKkGvX/Ab9T7P5+cvuRAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SvY8alfVMxFl",
        "colab_type": "code",
        "cellView": "both",
        "colab": {}
      },
      "source": [
        "#Guardar el modelo\n",
        "#@markdown optimizer_epochs_samples_features_accuracy_version\n",
        "\n",
        "nombre = \"LSTM_SGD_50e_650m_2f_0.86_V1\" #@param {type:\"string\"}\n",
        "model.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.h5')\n",
        "np.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.npy', le.classes_)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bCO3lpZXh5fA",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Cargar modelo\n",
        "model = tf.keras.models.load_model('/content/drive/My Drive/IA2/DATA/Models/model_LSTM_SGD_50e_650m_2f_0.86_V1.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "R6aRCV2ljH5h",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "66c74dbe-9914-4d70-ad8b-aef262c48d37"
      },
      "source": [
        "#@title Predicción\n",
        "X_test = x_test.reshape(x_test.shape[0],x_test.shape[1],1)\n",
        "y_predict = model.predict_classes(X_test)\n",
        "y_test_single = np.argmax(y_test, axis=1)\n",
        "conf_matrix = confusion_matrix(y_test_single, y_predict)\n",
        "print(conf_matrix)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[ 36   2   3   4  94   2]\n",
            " [  8  94  14  37  34   1]\n",
            " [ 16  17  31  19  83   2]\n",
            " [  8  33   3  69  41   8]\n",
            " [ 86  28  31  26 306  11]\n",
            " [ 17   4   3  13  28  88]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_zy5zOXUq_zD",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "ae0b70d4-7c24-4cfd-920d-7342b119732c"
      },
      "source": [
        "text = [\"ON\", \"OFF\", \"YES\", \"NO\", \"STOP\", \"UNKNOWN\"]\n",
        "#Vector para gráficas\n",
        "LSTM50=[]\n",
        "for i in range(len(text)):\n",
        "  acierto_promedio = conf_matrix[i][i]/(np.sum(conf_matrix[i], axis=0))\n",
        "  LSTM50.append(np.round(acierto_promedio, 2))\n",
        "  print(f\"Aciertos clase {text[i]}: {np.round(acierto_promedio*100, 2)}%\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Aciertos clase ON: 25.53%\n",
            "Aciertos clase OFF: 50.0%\n",
            "Aciertos clase YES: 18.45%\n",
            "Aciertos clase NO: 42.59%\n",
            "Aciertos clase STOP: 62.7%\n",
            "Aciertos clase UNKNOWN: 57.52%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "UYRZL_n5cdrz"
      },
      "source": [
        "### 100 Epochs"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "cellView": "both",
        "id": "WVngdRS1cdr4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "ebbb26f4-a02c-4044-9c05-b3292481134b"
      },
      "source": [
        "#@title Entrenamiento\n",
        "tf.keras.backend.clear_session()\n",
        "epochs = 100 #@param {type:\"slider\", min:50, max:2000, step:50}\n",
        "\n",
        "optimizer = tf.keras.optimizers.SGD(lr=1e-2, momentum=0.9)\n",
        "# optimizer = tf.keras.optimizers.Adam(lr=1e-4)\n",
        "# optimizer = tf.keras.optimizers.Adadelta(lr=1e-4)\n",
        "\n",
        "# optimizer = 'adam' #@param [\"adam\", \"SGD\", \"adadelta\"]\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              metrics=['accuracy'],\n",
        "              optimizer=optimizer)\n",
        "X = x_train.reshape(x_train.shape[0],x_train.shape[1],1)\n",
        "# print(X.shape, y_train.shape)\n",
        "history = model.fit(X, y_train,epochs=epochs,verbose=1)\n",
        "\n",
        "acc = history.history['accuracy']\n",
        "loss = history.history['loss']\n",
        "\n",
        "epochs=range(len(loss)) # Get number of epochs\n",
        "\n",
        "#------------------------------------------------\n",
        "# Plot acc and Loss\n",
        "#------------------------------------------------\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.plot(epochs, acc, 'red', label='acc')\n",
        "plt.title('Acc')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.show();\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.title('Loss')\n",
        "plt.plot(epochs, loss, 'black', label='loss')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Error\")\n",
        "plt.show();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.6584 - accuracy: 0.3710\n",
            "Epoch 2/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.5856 - accuracy: 0.3849\n",
            "Epoch 3/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.5104 - accuracy: 0.4136\n",
            "Epoch 4/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.4707 - accuracy: 0.4354\n",
            "Epoch 5/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.4374 - accuracy: 0.4431\n",
            "Epoch 6/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.4327 - accuracy: 0.4464\n",
            "Epoch 7/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.4138 - accuracy: 0.4482\n",
            "Epoch 8/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.3980 - accuracy: 0.4551\n",
            "Epoch 9/100\n",
            "122/122 [==============================] - 12s 94ms/step - loss: 1.3638 - accuracy: 0.4695\n",
            "Epoch 10/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.3630 - accuracy: 0.4731\n",
            "Epoch 11/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.3428 - accuracy: 0.4813\n",
            "Epoch 12/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.3214 - accuracy: 0.4851\n",
            "Epoch 13/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.3132 - accuracy: 0.4874\n",
            "Epoch 14/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.2757 - accuracy: 0.5092\n",
            "Epoch 15/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.2702 - accuracy: 0.5008\n",
            "Epoch 16/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.2478 - accuracy: 0.5146\n",
            "Epoch 17/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.2576 - accuracy: 0.5128\n",
            "Epoch 18/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.2278 - accuracy: 0.5197\n",
            "Epoch 19/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.2383 - accuracy: 0.5115\n",
            "Epoch 20/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.1973 - accuracy: 0.5305\n",
            "Epoch 21/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.1875 - accuracy: 0.5374\n",
            "Epoch 22/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.1940 - accuracy: 0.5285\n",
            "Epoch 23/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.1649 - accuracy: 0.5449\n",
            "Epoch 24/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.1561 - accuracy: 0.5531\n",
            "Epoch 25/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.1301 - accuracy: 0.5526\n",
            "Epoch 26/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.1204 - accuracy: 0.5590\n",
            "Epoch 27/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.0988 - accuracy: 0.5697\n",
            "Epoch 28/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.0982 - accuracy: 0.5615\n",
            "Epoch 29/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.0617 - accuracy: 0.5874\n",
            "Epoch 30/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.0461 - accuracy: 0.5867\n",
            "Epoch 31/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 1.0121 - accuracy: 0.6082\n",
            "Epoch 32/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 1.0038 - accuracy: 0.6023\n",
            "Epoch 33/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.9835 - accuracy: 0.6059\n",
            "Epoch 34/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.9397 - accuracy: 0.6359\n",
            "Epoch 35/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.9213 - accuracy: 0.6413\n",
            "Epoch 36/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.8801 - accuracy: 0.6577\n",
            "Epoch 37/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.8424 - accuracy: 0.6692\n",
            "Epoch 38/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.8228 - accuracy: 0.6751\n",
            "Epoch 39/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.8293 - accuracy: 0.6733\n",
            "Epoch 40/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.7608 - accuracy: 0.6926\n",
            "Epoch 41/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.7054 - accuracy: 0.7236\n",
            "Epoch 42/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.6863 - accuracy: 0.7292\n",
            "Epoch 43/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.6263 - accuracy: 0.7644\n",
            "Epoch 44/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.6295 - accuracy: 0.7564\n",
            "Epoch 45/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.5807 - accuracy: 0.7767\n",
            "Epoch 46/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.5407 - accuracy: 0.7956\n",
            "Epoch 47/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.5482 - accuracy: 0.7915\n",
            "Epoch 48/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.4428 - accuracy: 0.8274\n",
            "Epoch 49/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.4325 - accuracy: 0.8374\n",
            "Epoch 50/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.4260 - accuracy: 0.8387\n",
            "Epoch 51/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.3216 - accuracy: 0.8815\n",
            "Epoch 52/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.2706 - accuracy: 0.9044\n",
            "Epoch 53/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.2967 - accuracy: 0.8928\n",
            "Epoch 54/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.2518 - accuracy: 0.9141\n",
            "Epoch 55/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.2652 - accuracy: 0.9028\n",
            "Epoch 56/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.2750 - accuracy: 0.9005\n",
            "Epoch 57/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.3169 - accuracy: 0.8826\n",
            "Epoch 58/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.2680 - accuracy: 0.9033\n",
            "Epoch 59/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.1692 - accuracy: 0.9436\n",
            "Epoch 60/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.1063 - accuracy: 0.9690\n",
            "Epoch 61/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0924 - accuracy: 0.9749\n",
            "Epoch 62/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0876 - accuracy: 0.9759\n",
            "Epoch 63/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0986 - accuracy: 0.9677\n",
            "Epoch 64/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.1471 - accuracy: 0.9497\n",
            "Epoch 65/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.1616 - accuracy: 0.9436\n",
            "Epoch 66/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.1425 - accuracy: 0.9526\n",
            "Epoch 67/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.1186 - accuracy: 0.9585\n",
            "Epoch 68/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0800 - accuracy: 0.9756\n",
            "Epoch 69/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0477 - accuracy: 0.9877\n",
            "Epoch 70/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0652 - accuracy: 0.9813\n",
            "Epoch 71/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.1521 - accuracy: 0.9477\n",
            "Epoch 72/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.1706 - accuracy: 0.9418\n",
            "Epoch 73/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0909 - accuracy: 0.9703\n",
            "Epoch 74/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0342 - accuracy: 0.9915\n",
            "Epoch 75/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0174 - accuracy: 0.9967\n",
            "Epoch 76/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0459 - accuracy: 0.9879\n",
            "Epoch 77/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0796 - accuracy: 0.9738\n",
            "Epoch 78/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.1707 - accuracy: 0.9413\n",
            "Epoch 79/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.3085 - accuracy: 0.9013\n",
            "Epoch 80/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.1340 - accuracy: 0.9564\n",
            "Epoch 81/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0435 - accuracy: 0.9892\n",
            "Epoch 82/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0186 - accuracy: 0.9977\n",
            "Epoch 83/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0058 - accuracy: 0.9995\n",
            "Epoch 84/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.0025 - accuracy: 0.9997\n",
            "Epoch 85/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 0.0013 - accuracy: 1.0000\n",
            "Epoch 86/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 0.0011 - accuracy: 1.0000\n",
            "Epoch 87/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 9.1730e-04 - accuracy: 1.0000\n",
            "Epoch 88/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 8.1604e-04 - accuracy: 1.0000\n",
            "Epoch 89/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 7.3993e-04 - accuracy: 1.0000\n",
            "Epoch 90/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 6.7677e-04 - accuracy: 1.0000\n",
            "Epoch 91/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 6.2487e-04 - accuracy: 1.0000\n",
            "Epoch 92/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 5.8157e-04 - accuracy: 1.0000\n",
            "Epoch 93/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 5.4332e-04 - accuracy: 1.0000\n",
            "Epoch 94/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 5.1098e-04 - accuracy: 1.0000\n",
            "Epoch 95/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 4.8257e-04 - accuracy: 1.0000\n",
            "Epoch 96/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 4.5661e-04 - accuracy: 1.0000\n",
            "Epoch 97/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 4.3401e-04 - accuracy: 1.0000\n",
            "Epoch 98/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 4.1348e-04 - accuracy: 1.0000\n",
            "Epoch 99/100\n",
            "122/122 [==============================] - 11s 93ms/step - loss: 3.9479e-04 - accuracy: 1.0000\n",
            "Epoch 100/100\n",
            "122/122 [==============================] - 11s 94ms/step - loss: 3.7802e-04 - accuracy: 1.0000\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfgAAAGDCAYAAADHzQJ9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deZzNdf//8cdrxjb2tRJhSERZMsyEsl7KEkJdaF9/JdWUq6JcV8KVrqikTSqu1uuiBUNItrIVI0kMQrmSJTGYqQxm3r8/5ozvYIZZzpnPnDPP++12bs75fD7n83meczv1nM9uzjlEREQktIR5HUBERET8TwUvIiISglTwIiIiIUgFLyIiEoJU8CIiIiFIBS8iIhKCVPAiIiIhSAUvIqcxs5/MrLPXOUQk71TwIiIiIUgFLyI5YmYlzWy8me3yPcabWUnfuKpmNtvMDprZATNbamZhvnGPmdkvZpZkZpvNrJO3n0SkaCjmdQARCRpPADFAM8ABM4HhwN+BIcBOoJpv2hjAmVkDYDDQ0jm3y8zqAOEFG1ukaNIavIjk1A3ASOfcr865fcBTwE2+cceA6kBt59wx59xSl36ji1SgJNDIzIo7535yzm3zJL1IEaOCF5GcOh/Yken1Dt8wgLHAVmC+mW03s6EAzrmtQCwwAvjVzP5rZucjIgGngheRnNoF1M70upZvGM65JOfcEOdcXaAn8HDGvnbn3AfOuba+9zrgXwUbW6RoUsGLSHaKm1mpjAfwH2C4mVUzs6rAP4D3AMysh5ldaGYGHCJ903yamTUws46+g/GOAH8Cad58HJGiRQUvItmZQ3ohZzxKAfHAd8B64BtgtG/a+sACIBlYCbzqnFtM+v73Z4DfgD3AOcCwgvsIIkWXpR8HIyIiIqFEa/AiIiIhSAUvIiISglTwIiIiIUgFLyIiEoJU8CIiIiEopK5FX7VqVVenTh2vY4iIiBSINWvW/Oacq5bVuJAq+Dp16hAfH+91DBERkQJhZjuyG6dN9CIiIiFIBS8iIhKCVPAiIiIhSAUvIiISglTwIiIiIUgFLyIiEoJU8CIiIiFIBS8iIhKCVPAiIiIhSAUvIiISglTwIiIiIUgFnwXnHKtWreLbb7/1OoqIiEieqOCz4JyjZ8+ejBkzxusoIiIieaKCz0JYWBi9evVizpw5HDlyxOs4IiIiuaaCz8a1115LcnIyCxcu9DqKiIhIrqngs9GhQwfKlSvHjBkzvI4iIiKSayr4bJQsWZLu3bszc+ZMUlNTvY4jIiKSKyr4M+jduzf79u1j5cqVXkcRERHJFRX8GXTt2pUSJUowffp0r6OIiIjkigr+DMqXL0/nzp2ZMWMGzjmv44iIiOSYCv4sevfuzfbt21m/fr3XUURERHJMBX8WPXv2xMx0NL2IiAQVFfxZnHvuubRu3Vr74UVEJKio4HPg2muv5dtvv+XHH3/0OoqIiEiOqOBzoHfv3gDMnDnT4yQiIiI5o4LPgXr16nHppZdqM72IiAQNFXwO9erVi+XLl7N//36vo4iIiJyVCj6HevbsSWpqKnPnzvU6ioiIyFmp4HOoRYsWVK9enbi4OK+jiIiInJUKPofCwsK45pprmDdvHikpKV7HEREROSMVfC707NmTpKQkvvjiC6+jiIiInJEKPhc6duxI6dKltZleREQKPRV8LkRERNClSxfi4uJ08xkRESnUVPC51LNnT37++WfWrVvndRQREZFsqeBzqXv37piZrmonIiKFmgo+l8455xwuv/xy7YcXEZFCTQWfBz179uSbb75h586dXkcRERHJUsAK3swmm9mvZvZ9NuPbm9khM/vW9/hHpnFXm9lmM9tqZkMDlTGvevbsCcCsWbM8TiIiIpK1QK7B/xu4+izTLHXONfM9RgKYWTjwCtAVaAQMMLNGAcyZaw0bNuTCCy/k448/9jqKiIhIlgJW8M65L4EDeXhrK2Crc267c+4o8F+gl1/D5ZOZcfvtt7Nw4UIWLVrkdRwREZHTeL0P/nIzW2dmc82ssW9YDeDnTNPs9A0rVGJjY6lTpw6xsbEcP37c6zgiIiIn8bLgvwFqO+eaAi8BM/IyEzO728zizSx+3759fg14JhEREYwdO5b169fz5ptvFthyRUREcsKzgnfOHXbOJfuezwGKm1lV4BfggkyT1vQNy24+k5xzUc65qGrVqgU086n69u1Lu3btGD58OImJiQW6bBERkTPxrODN7DwzM9/zVr4s+4HVQH0zizSzEkB/oFCedG5mjB8/nsTERJ566imv44iIiJwQyNPk/gOsBBqY2U4zu8PM7jGze3yT9AO+N7N1wASgv0t3HBgMfAYkANOccxsClTO/mjVrxl133cUrr7xCQkKC13FEREQAsFC6aUpUVJSLj48v8OXu27eP+vXrEx0dzbx58/BtmBAREQkoM1vjnIvKapzXR9GHhGrVqjFq1Cjmz5/P5MmTvY4jIiKigveX++67j44dOxIbG8v27du9jiMiIkWcCt5PwsLCmDJlCmFhYdx6662kpqZ6HUlERIowFbwf1apVi5deeomlS5fywgsveB1HRESKMBW8n910001ce+21PPHEE6xfv97rOCIiUkSp4P3MzHj99depWLEivXr14vHHH+eTTz7h559/JpTOWBARkcJNBR8A1apVY+rUqVSqVImxY8fSt29fatWqRWRkJMuWLfM6noiIFAEq+ABp3749a9asISkpia+//pqXX36Z4sWL07FjR51KJyIiAaeCD7BSpUrRqlUr7rvvPlatWkW7du244447eOihh3QXOhERCRgVfAGqVKkSc+fO5YEHHmD8+PF0796dgrwDnoiIFB0q+AJWrFgxXnzxRSZNmsSSJUu49NJLmT17ttexREQkxKjgPXLXXXexevVqzj33XK655hruuusukpKSvI4lIiIhQgXvoSZNmrBq1SqGDRvG5MmTadq0Kdu2bfM6loiIhAAVvMdKlizJ008/zZdffkliYiK33HKLLnMrIiL5poIvJNq0acOECRNYvnw5L774otdxREQkyKngC5Ebb7yRa665hieeeILNmzd7HUdERIKYCr4QybjMbUREBLfddps21YuISJ6p4AuZ6tWr89JLL7Fy5UrdkU5ERPKsmNcB5HQDBw7kww8/ZPjw4ezYsYPSpUtTqlQpypQpw/XXX0+dOnW8jigiIoWchdIdzqKiolx8fLzXMfxiz5499OjRg23btnHkyBGOHDkCQNWqVZk9ezbR0dEeJxQREa+Z2RrnXFRW47SJvpA677zziI+PJzExkT///JO0tDQSEhIoX748HTp0IC4uzuuIIiJSiKngg4SZ0bBhQ1asWEHjxo259tprmThxotexRESkkFLBB5lzzz2XJUuW0LVrV+69916GDx9OKO1mERER/1DBB6EyZcowY8YM7rjjDv75z38yePBg0tLSvI4lIiKFiI6iD1LFihXjjTfeoFKlSowbN45Dhw4xZcoUihcv7nU0EREpBFTwQczMePbZZ6lcuTKPP/44hw8fZurUqURERHgdTUREPKaCD3JmxrBhw6hYsSL33XcfjRo14sYbb+SGG26gYcOGXscTERGPaB98iLj33nuZPXs29evX5+mnn+biiy+mRYsWfPTRR15HExERD6jgQ0i3bt2YP38+O3fuZPz48aSkpHDDDTfoxjUiIkWQCj4EVa9enQcffJCFCxcSERHB4MGDdSqdiEgRo4IPYeeeey5PP/00CxYsYNq0aV7HERGRAqRr0Ye41NRUoqOj2bVrF5s2baJ8+fJeRxIRET/RteiLsPDwcF577TX27NnDP/7xD6/jiIhIAVHBFwEtW7bknnvu4aWXXuLbb7/1Oo6IiBQAbaIvIhITE2nYsCERERFcd911tG3blrZt21KlShWvo4mISB5pE71QqVIlPvjgA2rWrMmECRPo3bs3VatWpU2bNuzevdvreCIi4mcBK3gzm2xmv5rZ99mMv8HMvjOz9Wa2wsyaZhr3k2/4t2amVXI/6dSpE8uWLePQoUN8+eWXjBo1inXr1tGuXTt27tzpdTwREfGjQK7B/xu4+gzjfwTaOecuBUYBk04Z38E51yy7TQ+Sd6VKleKKK65g+PDhzJ8/nz179tCuXTt27NhxYhrnHAsXLmT48OEcPnzYw7QiIpIXAbsWvXPuSzOrc4bxKzK9/AqoGagskr3WrVuzYMECunTpQrt27YiLi2PZsmW8/PLLJCQkALB582amTZuGmXmcVkREcqqw7IO/A5ib6bUD5pvZGjO726NMRUarVq1YtGgRSUlJNG3alPvuu48yZcrw9ttvM3r0aD766CPGjx/vdUwREckFz+8mZ2YdSC/4tpkGt3XO/WJm5wCfm9km59yX2bz/buBugFq1agU8b6i67LLLWLJkCW+99RYDBgwgOjoaSN9UHx8fz6OPPkqrVq1o06bNSe/bv38/lStX1tq9iEgh4+kavJk1Ad4Eejnn9mcMd8794vv3V2A60Cq7eTjnJjnnopxzUdWqVQt05JB26aWXMn78+BPlDum3o50yZQq1a9fm+uuv59dff8U5x5IlS+jevTtVq1bV2r2ISCHkWcGbWS3gE+Am59yWTMPLmFm5jOdAFyDLI/GlYFSsWJGPP/6YAwcO0LNnT1q1akWHDh1YvXo1jRs35qmnnuK3337zOqaIiGQSyNPk/gOsBBqY2U4zu8PM7jGze3yT/AOoArx6yulw5wLLzGwdsAr41Dk3L1A5JWeaNm3Ka6+9xtdff82hQ4eYOHEiO3bs4MMPPyQ5OZmnnnrK64giIpKJrmQnubJ9+3Zq165NeHj4iWGDBg1i0qRJfP/99zRs2NDDdCIiRYuuZCd+U7du3ZPKHWDEiBGULl2aRx991KNUIiJyKhW85Ns555zD448/zqxZs1i8eLHXcUREBBW8+ElsbCy1atViyJAhpKWleR1HRKTIU8GLX5QqVYoxY8awdu1a7rzzTrZs2ZLldD/99BOJiYkFnE5EpOhRwYvf9O/fn0GDBvHee+/RoEEDunXrxpw5c5g7dy4PPPAA9evXJzIyktatW5OUlOR1XBGRkKaj6MXv9uzZw6RJk3jttdfYs2cPkL6G36FDB1q0aMGYMWPo1asXH330ka6AJyKSD2c6il4FLwFz9OhRPv30UyIiImjXrh0REREAPPfcc/ztb3/jmWee4bHHHvM4pYhI8FLBS6HinKN///589NFHfPbZZ3Tu3NnrSCIiQUnnwUuhYma89dZbXHzxxfTv3/+k+9CLiIh/qODFE2XLlmX69OkcO3aM6Ohohg0blu2R9yIiknsqePFM/fr1mT9/Pq1atWLs2LE0aNCAK6+8kpkzZ3odTUQk6KngxVPR0dHExcXxv//9jzFjxrB792569+7NtGnTvI4mIhLUdJCdFCp//vknf/nLX1i9ejULFizgiiuu8DqSiEihpYPsJGhEREQwc+ZMIiMj6dWrFwkJCV5HEhEJSip4KXSqVKnC3LlzKVGiBF27dmX37t1eRxIRCToqeCmUIiMj+fTTT/ntt9/o2LEjU6dO5dixY17HEhEJGip4KbRatGjBjBkzOHr0KP3796d27dqMHDnyxOVvRUQkeyp4KdQ6d+7MDz/8wOzZs2nSpAlPPvkk9erVY+7cuV5HExEp1FTwUuiFhYXRvXt35s2bx6ZNm2jQoAG9evXiww8/9DqaiEihpYKXoNKgQQMWLVpEq1at6N+/P1OmTPE6kohIoVTM6wAiuVWxYkU+++wz+vTpw+23386+ffu4+uqrKVGiBCVKlKBChQpUqVLF65giIp7ShW4kaKWkpDBgwACmT59+0vDw8HCmTp1K3759PUomIlIwdLtYCVnHjx9nyZIlHD58mKNHj3L06FFeeOEFdu/eTUJCApUqVfI6oohIwJyp4LWJXoJasWLFTruf/CWXXELLli0ZNmwYEydO9CiZiIi3dJCdhJzLLruM2NhYXn/9dZYvX+51HBERT2gTvYSk5ORkGjduTNmyZVm7di0lSpTwOpKIiN/pZjNS5JQtW5ZXXnmFjRs3MnbsWK/jiIgUOBW8hKwePXrQr18/Ro0axZo1a7yOIyJSoFTwEtImTJhA5cqVufzyyxkzZgzHjx/3OpKISIFQwUtIq169Ot999x29e/fm8ccf54orrmDLli1exxIRCTgVvIS8qlWrMnXqVD744AM2b95Ms2bNmD17ttexREQCSgUvRYKZMWDAAL7//nsaNGjArbfeyt69e72OJSISMCp4KVLOP/983n//fZKTk7nnnnsIpdNERUQyU8FLkdOoUSNGjRrFjBkz+OCDD7yOIyISECp4KZIefvhhWrduzeDBg9m1a5fXcURE/E4FL0VSeHg4U6ZMISUlhbvuukub6kUk5AS04M1sspn9ambfZzPezGyCmW01s+/M7LJM424xsx98j1sCmVOKposuuogxY8YwZ84cBg8eTEJCgteRRET8JtBr8P8Grj7D+K5Afd/jbuA1ADOrDDwJRAOtgCfNTPf9FL+7//77ueWWW3j99ddp1KgRLVu2ZMKECSQmJnodTUQkXwJa8M65L4EDZ5ikF/COS/cVUNHMqgNXAZ875w445xKBzznzHwoieRIWFsa///1vdu7cyfPPP8/x48d58MEHad68udboRSSoeb0Pvgbwc6bXO33Dsht+GjO728zizSx+3759AQsqoe28887joYceYu3atSxfvpwjR47Qpk0bli1b5nU0EZE88brg8805N8k5F+Wci6pWrZrXcSQEtG7dmpUrV3LOOefQuXNnPvroI68jiYjkmtcF/wtwQabXNX3DshsuUiAiIyNZvnw5LVq04Prrr+f111/3OpKISK54XfBxwM2+o+ljgEPOud3AZ0AXM6vkO7iui2+YSIGpUqUKCxYsoEuXLsTGxup8eREJKoE+Te4/wEqggZntNLM7zOweM7vHN8kcYDuwFXgDGATgnDsAjAJW+x4jfcNEClRERASvvvoqx48fZ/To0V7HERHJMQulC3xERUW5+Ph4r2NICBo0aBBvvPEGmzdvpm7dul7HEREBwMzWOOeishrn9SZ6kaAwfPhwihcvzogRI7yOIiKSIyp4kRw4//zzuf/++3nvvffYsGGD13FERM5KBS+SQ48++ijlypXj73//u9dRRETOSgUvkkNVqlThb3/7G9OnT2f16tVexxEROSMVvEguxMbGUrVqVR588EGSkpK8jiMiki0VvEgulCtXjhdffJFVq1YRExPDDz/84HUkEZEsqeBFcmngwIHMnz+fvXv30rJlS+bOnet1JBGR06jgRfKgY8eOxMfHU6dOHbp3787YsWO9jiQichIVvEge1alThxUrVtC3b18effRRVq5c6XUkEZETVPAi+VC6dGmmTJnCeeedx5AhQwilK0OKSHBTwYvkU9myZRk1ahQrV67k448/9jqOiAiga9GL+EVqairNmjXjjz/+ICEhgRIlSngdSUSKAF2LXiTAwsPDGTduHNu3b+eVV17xOo6IiApexF+uuuoqunTpwqhRozhwQHc3FhFvnbXgzSzMzFoXRBiRYDd27FgOHjzIP//5T6+jiEgRd9aCd86lAdrmKJIDTZo04fbbb+ell15iy5YtXscRkSIsp5voF5pZXzOzgKYRCQGjR4+mdOnS3HfffTptTkQ8k9OC/3/Ah8BRMztsZklmdjiAuUSC1nnnncc///lPFixYwIcffuh1HBEponSanEgApKam0qpVK3bv3s2mTZsoX76815FEJAT55TQ5M+tpZuN8jx7+iycSesLDw3n11VfZs2cPI0aM8DqOiBRBOSp4M3sGeBDY6Hs8aGZjAhlMJNhFR0dz9913M2HCBNatW+d1HBEpYnK0id7MvgOa+Y6ox8zCgbXOuSYBzpcr2kQvhc2BAwdo0KAB9evXZ+nSpYSHh3sdSURCiL+uZFcx0/MK+YskUjRUrlyZ5557jpUrV9KvXz9+//13ryOJSBFRLIfTPQ2sNbPFgAFXAkMDlkokhNx8880cPHiQ2NhY2rVrx6xZs6hevbrXsUQkxOXoSnZAGhADfAJ8DFzunJsa4GwiIeOBBx5g5syZbNq0iejoaL777jsSEhJ4/vnn6dKlC+XLl2fOnDlexxSREJLTffDx2W3jL0y0D14Ku7Vr19KjRw9279594iI4F198Mbt376Z9+/ZMnz7d44QiEkz8sQ9+gZn9zcwuMLPKGQ8/ZhQpEpo3b87XX3/NQw89xMSJE/nxxx/ZuHEjN998M3PnziUpKcnriCISInK6Bv9jFoOdc66u/yPlndbgJVgtX76ctm3b8v777zNw4ECv44hIkMjXGrxvH/xQ51zkKY9CVe4iwezyyy+nRo0aTJs2zesoIhIicno3uUcKIItIkRUWFka/fv2YN28ehw/rNg8ikn/aBy9SSFx//fWkpKQwa9Ysr6OISAjIacH/FbgP+BJY43toZ7eIH8XExFCjRg3dgU5E/CJHF7pxzkUGOohIURcWFsZ1113Hq6++yuHDh3UHOhHJlzOuwZvZo5meX3fKuKcDFUqkqLruuus4evQocXFxXkcRkSB3tk30/TM9H3bKuKv9nEWkyIuJiaFmzZraTC8i+Xa2grdsnmf1+vQ3m11tZpvNbKuZnXbtejN7wcy+9T22mNnBTONSM43T6owUCRmb6efNm8ehQ4e8jiMiQexsBe+yeZ7V65P4bin7CtAVaAQMMLNGJ83AuYecc82cc82Al0i/1n2GPzPGOed6niWnSMjI2Ew/c+ZMr6OISBA7W8E3NbPDZpYENPE9z3h96Vne2wrY6pzb7pw7CvwX6HWG6QcA/8lxcpEQFR0dTb169XjzzTe9jiIiQeyMBe+cC3fOlXfOlXPOFfM9z3hd/CzzrgH8nOn1Tt+w05hZbSASWJRpcCkzizezr8ysdw4+i0hICAsL45577mHp0qWsX7/e6zgiEqRyeh58oPUHPnLOpWYaVtt3fd2BwHgzq5fVG83sbt8fAvH79u0riKwiAXfbbbdRsmRJJk6c6HUUEQlSgSz4X4ALMr2u6RuWlf6csnneOfeL79/twBKgeVZvdM5Ncs5FOeeiqlWrlt/MIoVClSpV6N+/P++8847uMCcieRLIgl8N1DezSDMrQXqJn3Y0vJk1BCoBKzMNq2RmJX3PqwJtgI0BzCpS6AwaNIjk5GTee+89r6OISBAKWME7544Dg4HPgARgmnNug5mNNLPMR8X3B/7rTr5v7cVAvJmtAxYDzzjnVPBSpLRs2ZIWLVrw6quvkpPbOouIZJaj+8EHC90PXkLNW2+9xZ133smXX37JFVdc4XUcESlk8nU/eBHxzoABA6hQoQKvvvqq11FEJMio4EUKsdKlS3Pbbbfx8ccfs3fvXq/jiEgQUcGLFHL33HMPx44d4x//+AdpaWlexxGRIKGCFynkGjRoQGxsLJMmTaJPnz4kJyd7HUlEgoAKXiQIPP/887z44ovMmjWLNm3asGPHDq8jiUghp4IXCQJmxgMPPMDcuXPZsWMHLVu2ZNWqVV7HEpFCTAUvEkS6dOnCV199RUREBHfffbfXcUSkEFPBiwSZhg0b8uijj7Ju3TrWrl3rdRwRKaRU8CJBaMCAAZQsWZLJkyd7HUVECikVvEgQqly5Mtdeey3vv/8+R44c8TqOiBRCKniRIHX77beTmJhIXNxp93ASEVHBiwSrjh07UqtWLW2mF5EsqeBFglR4eDi33nor8+fP5+eff/Y6jogUMip4kSB266234pzj7bff9jqKiBQyKniRIBYZGUmHDh2YMmWKrlMvIidRwYsEudtvv53t27ezdOlSABITE1m9ejXbtm3zOJmIeMmcc15n8JuoqCgXHx/vdQyRAvXHH39QvXp1SpUqxbFjx0hMTATST6XbsWMHZcuW9TihiASKma1xzkVlNU5r8CJBrnTp0owePZqmTZvSv39/nnvuOV5++WUOHDjApEmTvI4nIh7RGrxIiOrYsSObN29m+/btlCxZ0us4IhIAWoMXKYKGDh3Krl27eP/9972OIiIeUMGLhKi//OUvNG/enGeffZbU1FSv44hIAVPBi4QoM2Po0KFs3ryZmTNneh1HRAqYCl4khPXt25d69erxzDPPEErH24jI2angRUJYeHg4jz76KKtXr2bRokVexxGRAqSCFwlxt9xyC9WrV2fMmDFeRxGRAqSCFwlxJUuWZMiQISxcuJDPP//c6zgiUkB0HrxIEZCSkkKjRo2IiIjg22+/pVixYl5HEhE/0HnwIkVcyZIlGTduHBs2bOCNN97wOo6IFAAVvEgR0bt3b9q3b8/f//73E9erF5HQpYIXKSLMjPHjx5OYmMjIkSO9jiMiAaaCFylCmjZtyp133snLL7/M5s2bAdi0aROjR49m4MCBJCcne5xQRPxFB9mJFDG//vor9evXJzIyktTUVL7//vsT4+Li4rjmmms8TCciuaGD7ETkhHPOOYdRo0bx3XffUblyZSZMmMAPP/xAeHg4X331ldfxRMRPtAYvUgQ55/jzzz8pXbr0iWEtWrSgYsWKLFy40MNkIpIbWoMXkZOY2UnlDhATE8Pq1at15zmREKGCFxEgveCTkpJISEjwOoqI+EFAC97MrjazzWa21cyGZjH+VjPbZ2bf+h53Zhp3i5n94HvcEsicIgLR0dEA2g8vEiICVvBmFg68AnQFGgEDzKxRFpNOdc418z3e9L23MvAkEA20Ap40s0qByioiUL9+fSpVqsTXX3/tdRQR8YNArsG3ArY657Y7544C/wV65fC9VwGfO+cOOOcSgc+BqwOUU0RI3y8fExOjNXiREBHIgq8B/Jzp9U7fsFP1NbPvzOwjM7sgl+8VET+Kjo5mw4YNHD582OsoIpJPXh9kNwuo45xrQvpa+tu5nYGZ3W1m8WYWv2/fPr8HFClKYmJicM6h001Fgl8gC/4X4IJMr2v6hp3gnNvvnEvxvXwTaJHT92aaxyTnXJRzLqpatWp+CS5SVLVq1QrQgXYioSCQBb8aqG9mkWZWAugPxGWewMyqZ3rZE8g4P+czoIuZVfIdXNfFN0xEAqhSpUo0aNBABS8SAooFasbOueNmNpj0Yg4HJjvnNpjZSCDeOW2PfIcAABv0SURBVBcHPGBmPYHjwAHgVt97D5jZKNL/SAAY6Zw7EKisIvJ/YmJimDNnDs45zMzrOCKSR7pUrYicZOLEidx7771s376dyMhIr+OIyBnoUrUikmO64I1IaFDBi8hJLr30UiIiIlTwIkFOBS8iJylWrBgtW7bUFe1EgpwKXkROEx0dzdq1a0lJSTn7xCJSKKngReQ0MTExHD16lCFDhrBz506v44hIHqjgReQ03bp1Y+DAgbz22mtERkZy4403snbtWq9jiUguqOBF5DSlSpXi/fffZ9u2bQwePJiZM2dy2WWX0bBhQx566CHmz5/PkSNHvI4pImeg8+BF5KwOHjzIe++9x+zZs1myZAkpKSmULl2auLg4OnXq5HU8kSJL58GLSL5UrFiRwYMHM2/ePA4cOMCnn35KREQEb7+d6/tDiUgBUcGLSK6ULl2abt26ccUVV+hceZFCTAUvInkSExPDDz/8wP79+72OIiJZUMGLSJ7ExMQA6II4IoWUCl5E8iQqKorw8HBtphcppFTwIpInZcqUoUmTJqxcudLrKCKSBRW8iORZTEwMX3/9NampqV5HEZFTqOBFJM9iYmJISkpi06ZNXkcRkVOo4EUkzzIOtNN+eJHCRwUvInlWv359KleurIIXKYRU8CKSZ2ZGTEyMCl6kEFLBi0i+xMTEsGHDBg4dOuR1FBHJRAUvIvkSExODc47Vq1d7HUVEMlHBi0i+tGrVCjPTZnqRQkYFLyL5UqFCBRo1aqSCFylkVPAikm8ZB9o557yOIiI+KngRybeYmBj279/P1q1bvY4iIj4qeBHJN13wRqTwUcGLSL5dfPHFVKhQgc8++yxgyzh69Cjz58/XbgCRHFLBi0i+hYeHc8sttzBt2jR27doVkGWMHj2aq666ilmzZgVk/iKhRgUvIn4RGxtLamoqL730kt/nvWvXLp577jkAxowZo7X4ADp8+DAdO3Zkw4YNXkeRfFLBi4hfREZG0qdPHyZOnEhycrJf5z1ixAiOHTvGww8/zFdffcXSpUv9On/5P6tWrWLx4sV8/vnnXkeRfFLBi4jfDBkyhIMHDzJ58uTTxo0bN47nn38+1/PcuHEjb731FoMGDWL06NFUq1aNZ555xh9xJQsbN24EYPv27R4nkfxSwYuI38TExNC6dWvGjx9PamrqieHjx4/nkUce4ZFHHmHLli25muewYcMoW7Ysw4cPJyIiggcffJC5c+eybt06f8cXOLFp/scff/Q4ieSXCl5E/GrIkCH8+OOPTJ8+HYBPPvmEhx9+mK5du1KqVClGjhyZ43l9+eWXxMXFMWzYMKpWrQrAoEGDKFu2LP/6178Ckr+o0xp86LBQOlglKirKxcfHex1DpEhLTU2lQYMGVKtWjeeee45OnTrRrFkzFi1axIgRIxg7dizff/89jRo1OuN8nHPExMSwa9cutmzZQkRExIlxjzzyCM8//zw//PADdevWDfRHKjKcc1SpUoXExEQiIiL4/fffMTOvY8kZmNka51xUVuO0Bi8ifhUeHk5sbCxfffUVV111FTVr1iQuLo6IiAgeeeQRypQpw4gRI844j/Xr13PrrbeyatUqRo0adVK5Azz00EMUK1aMcePGBfCTFD179uwhMTGRBg0a8Oeff7J3716vI0k+qOBFxO9uu+02KlWqRKlSpZg7dy7VqlUDoGrVqjz44IN8+OGHfPfddye95/jx43z88ce0b9+eJk2a8OGHH3L//fdz0003nTb/888/n5tvvpnJkyerhPwoY/N8jx49AG2mD3YBLXgzu9rMNpvZVjMbmsX4h81so5l9Z2YLzax2pnGpZvat7xEXyJwi4l9lypThiy++YNWqVVx44YUnjRsyZAjly5fnySefPDFs0aJFNG/enH79+vHTTz/x7LPPsnPnTiZMmEB4eHiWyxg8eDApKSnMnTs3oJ+lKMk4wC6j4HWgXXALWMGbWTjwCtAVaAQMMLNTd7qtBaKcc02Aj4BnM4370znXzPfoGaicIhIYl156KZGRkacNr1SpEg8//DAzZsxg+vTpXHfddXTq1Ink5GSmTp3Ktm3beOSRR6hcufJZ51+xYkWWLVsWqI9Q5GzYsIFKlSqduLeA1uCDWyDX4FsBW51z251zR4H/Ar0yT+CcW+yc+8P38iugZgDziEghERsbS6VKlejTpw+ffvopo0aNYuPGjVx//fXZrrGfKiwsjDZt2rB8+fIApy06Nm7cSOPGjSlVqhQ1atRQwQe5QBZ8DeDnTK93+oZl5w4g87a2UmYWb2ZfmVnv7N5kZnf7povft29f/hKLSIGoUKECr7zyCnfddRebN28+cY57brVp04ZNmzbx22+/BSBl0eKcY8OGDTRu3BhIvzKhNtEHt0JxkJ2Z3QhEAWMzDa7tO/R/IDDezOpl9V7n3CTnXJRzLirjQB4RKfwGDBjApEmTuOCCC/I8j7Zt2wKwYsUKf8Uqsvbu3UtiYuKJ0xfr1q2rNfggF8iC/wXI/F9uTd+wk5hZZ+AJoKdzLiVjuHPuF9+/24ElQPMAZhWRIBQVFUXx4sW1md4PMg6wy7wGv3PnTlJSUs70NinEAlnwq4H6ZhZpZiWA/sBJR8ObWXPgddLL/ddMwyuZWUnf86pAG2BjALOKSBCKiIigRYsWOtDODzIKPvMavHOO//3vf17GknwIWME7544Dg4HPgARgmnNug5mNNLOMo+LHAmWBD085He5iIN7M1gGLgWeccyp4ETlN27ZtiY+P58iRI15HCWobN26kUqVKnHfeeQAnrhCozfTBq1ggZ+6cmwPMOWXYPzI975zN+1YAlwYym4iEhjZt2jBu3Dji4+NP7JOX3Ms4wC7j0rQZpziq4INXoTjITkQkr9q0aQOg/fD5kHEEfeb7A1SvXp2SJUvqSPogpoIXkaBWrVo1LrroIhV8PmQcQZ9xgB2kX2cgMjJSa/BBTAUvIkGvbdu2LF++nLS0NK+jBKVTD7DLoHPhg5sKXkSCXps2bThw4ACbN2/2OkpQyrjJTOY1eNC58MFOBS8iQS9jP7xOl8ubjGvQZxxBnyEyMpKDBw+SmJjoUTLJDxW8iAS9iy66iKpVq2o/fB5lHGCXcQR9hoxT5bSZPjip4EUk6JkZbdq00Rp8Hpx6DfrMdC58cFPBi0hIaNu2Ldu2bWPv3r1eR/GrxMREfv/994DN/9Rr0Gemc+GDmwpeREJCxn74L774wuMk/jNjxgxq1arFXXfdFbBlrFu3DoBLLrnktHHly5enSpUq2kQfpFTwIhISWrZsyTnnnMPUqVMDtow5c+bwr3/9i2PHjgVsGQBpaWmMHDmSa6+9lpSUFGbNmsXRo0cDsqwvvviC8PBwoqOjsxyvI+mDlwpeREJCsWLFGDhwILNmzeLAgQN+nffu3bu5/vrr6d69O0OHDuWqq65i//79fl1GhqSkJPr168eTTz7JzTffzLvvvktycnLADiBcsmQJLVu2pGzZslmO18VugpcKXkRCxk033cSxY8eYNm2aX+aXlpbGxIkTufjii4mLi2P06NFMnjyZ5cuXEx0dTUJCgl+Wk8E5R5cuXZg5cyYvvPAC//73v+nevTvFixdnzpw5Z59BLiUnJ7N69Wo6dOiQ7TR169Zlx44dpKam+n35ElgqeBEJGc2bN6dx48a8++67fplfbGws9957Ly1atGD9+vU88cQT3HbbbSxZsoSkpCRiYmKYN2+eX5YFsHXrVr766iueffZZYmNjMTPKli3LlVdeydy5c/22nAzLly/n+PHjtG/fPttpIiMjOXbsGL/88ovfly+BpYIXkZBhZtx0002sWLGCrVu35mteO3fuZOLEidxxxx0sWLCA+vXrnxh3+eWXs2rVKiIjI+nRowfTp0/Pb3QAFi5cCEDPnj1PGt6tWzc2bNjg93uzL1myhGLFitG6detsp9GpcsFLBS8iIeWGG27AzHjvvffyNZ/nn3+etLQ0hg8fftoFYABq167N0qVLadmyJX/961/9sia/YMECLrjgAi688MKThnfr1g3A72vxixcvplWrVtnufwcVfDBTwYtISKlZsyYdO3bk3XffxTmXp3ns37+fSZMmMXDgQOrUqZPtdOXKlWPu3LlccsklXHvttSxZsiRvoYHU1FQWL15M586dT/uDokGDBtSpU8ev++GTkpKIj48/4+Z5gFq1alG2bFnWrFnjt2VLwVDBi0jIuemmm9i+fTsrVqzI0/tfeuklfv/9dx577LGzTluxYkXmz59P3bp16dGjBytXrszTMr/99lsOHDhA586dTxtnZnTr1o2FCxeSkpKSp/mfavny5aSmpp7xADtIPzuhbdu2LF682C/LlYKjgheRkNOnTx8iIiLydLBdcnIyEyZMoFevXllevjUrVatWZcGCBVSvXp2uXbvy008/5Xq5GfvfO3bsmOX4rl278vvvv7N06dJczzsrixcvpnjx4mfc/56hQ4cOJCQksGfPHr8sWwqGCl5EQk65cuXo06cPU6dOzfUa76RJk0hMTGTYsGG5el/16tWZP38+x44dY8iQIbl6L6Tvf7/kkktOu6Nbhg4dOlCyZEm/baZfsmQJ0dHRlC5d+qzTZmzGD6WrBBYFKngRCUk33XQTBw8eZPz48aSlpeXoPSkpKTz33HN06NAh2yu7nUlkZCTDhw/nk08+Yf78+Tl+35EjR1i2bBmdOnXKdpoyZcrQvn17vxxod/jwYdasWXPW/e8ZLrvsMsqVK6fN9EFGBS8iIalz58506tSJoUOH0q5dO9avX3/W90yePJldu3bleu09s4cffpgLL7yQ+++/P8eXl125ciV//vlnlvvfM+vatSubNm3K9xHty5Yty9H+9wzFihXjyiuvVMEHGRW8iISk8PBw5s+fz+TJk0lISKB58+YMGTIkywu2rFixgu7duzNo0CBiYmLOWrRnUrJkSSZMmMCWLVsYP358jt6zYMECwsPDufLKK884nb9Ol1uyZAklSpQgJiYmx+/p0KEDW7ZsYdeuXflathQcy+tpJIVRVFSUi4+P9zqGiBQy+/fv5/HHH2fSpElA+v7yli1bctlll7FkyRKWLFlClSpViI2N5f7776dChQr5XmavXr1YuHAhmzdvpkaNGmecNiYmhvDw8Bxdb/7CCy/kwgsvzNd59y1btiQiIoIvv/wyx+/55ptvaNGiBe+99x433HBDnpct/mVma5xzUVmN0xq8iIS8KlWq8Prrr7N+/XpefPFFOnXqxObNmxkxYgRbtmzh+eefZ8eOHQwfPtwv5Q7wwgsvcPz4cR555JEzTnfw4EFWr159xv3vmQ0YMIDPP/88z1e1O3ToEN98802ON89naNq0KRUrVtRm+iCigheRIuOSSy7hgQce4N1332XTpk0cPnyYn376iYceeogyZcr4dVl169blscce4z//+c8ZD7j74osvSEtLy/FugTvvvBPnHG+++Waecs2ZM4e0tLQcH2CXITw8nHbt2uXrYj5SsFTwIlJklStXjuLFiwds/kOHDqVRo0bceuut/Pbbb1lOs2DBAkqXLp3j/eG1a9emW7duvPnmm7m+L/3+/ft5+OGHadKkCW3bts3VeyH9dLlt27bx888/5/q9UvBU8CIiARIREcH777/P/v37T6x5Z+acY8GCBVx55ZWUKFEix/P9f//v/7F7925mz56dqzz33Xcf+/fv55133snTHzYZm/W1mT44qOBFRAKoWbNmjBkzhpkzZ560WT05OZkbb7yRTZs2cc011+Rqnl27dqVmzZpMnDgxx++ZOnUqU6dO5cknn6Rp06a5Wl6GSy+9lCpVqqjgg4SOohcRCbC0tDSuuuoqVqxYwTfffENqair9+vVj8+bNjBo1iqFDhxIWlrv1rZEjR/Lkk0+ybdu2E3d8y86ePXto3Lgx9erVY8WKFRQrVizPn6Vv376sWbMmT5fjFf/TUfQiIh4KCwvj7bffplSpUvTo0YOWLVuyf/9+FixYwOOPP57rcge44447CA8PP3HqX3acc9x111388ccfvPPOO/kqd0jfTL9jxw5+/PHHfM1HAk8FLyJSAM4//3zefPNNtm7dSosWLVi7dm2uT1XLrEaNGlxzzTVMnjz5jFfMmzRpErNnz2bMmDE0bNgwz8vLkJF50aJF+Z6XBJY20YuIFKAffviByMjIfK9JA8ybN4+uXbvy3//+l7/+9a+njV+xYgXt27enQ4cOzJ07N09bCk7lnOOiiy4iOTmZZcuWUa9evXzPU/LuTJvoVfAiIkEqLS2NevXqUbFiRT777DPOOeecE+N27dpFixYtKFOmDKtWraJy5cp+W25CQgJXXHEFFSpUYNmyZVSvXt1v85bc0T54EZEQFBYWxrhx40hISKBp06Yn7imfkpJCnz59SEpKYsaMGX4td4CLL76YOXPmsHfvXq666ioSExP9On/xDxW8iEgQ69u3L6tWraJixYr85S9/4YknnmDQoEF8/fXXvPPOO1xyySUBWW6rVq2YMWMGmzZtokePHvzxxx8BWY7knQpeRCTINWnShPj4eG677TaefvppJk+ezPDhw+nTp09Al9u5c2c++OADVq5cSevWrXXgXSET0II3s6vNbLOZbTWzoVmML2lmU33jvzazOpnGDfMN32xmVwUyp4hIsCtTpgxvvfUWU6dOZfjw4Tz11FMFstx+/foxffp0Dh48SKdOnbjmmmvYtGlTgSxbzixgB9mZWTiwBfgLsBNYDQxwzm3MNM0goIlz7h4z6w9c65z7q5k1Av4DtALOBxYAFznnUs+0TB1kJyLijSNHjvDiiy/y9NNP8/vvv9O1a1caNmzIRRddRP369bngggsoX7485cqVo2TJkpiZ15FDgidH0ZvZ5cAI59xVvtfDAJxzYzJN85lvmpVmVgzYA1QDhmaeNvN0Z1qmCl5ExFv79u1j9OjRLFiwgG3btpGSknLaNMWLF6d06dKEh4cTFhZGeHg44eHhmBlmRlhY2InnOXkApz3Pzb+ZnW2a/L6OjY3N8pTGvDpTwef/RMzs1QAy33JoJxCd3TTOueNmdgio4hv+1SnvrZHVQszsbuBugFq1avkluIiI5E21atV48cUXAUhNTeXnn39my5Yt7Nq1i6SkpBOPP/74g7S0NFJTU088nHMnHmlpaSe9zu4BnPY8N/9mdrZp8vsayNVNhfIrkAVfIJxzk4BJkL4G73EcERHxCQ8Pp06dOtSpU8frKEVSIA+y+wW4INPrmr5hWU7j20RfAdifw/eKiIhINgJZ8KuB+mYWaWYlgP5A3CnTxAG3+J73Axa59G0acUB/31H2kUB9YFUAs4qIiISUgG2i9+1THwx8BoQDk51zG8xsJBDvnIsD3gLeNbOtwAHS/wjAN900YCNwHLjvbEfQi4iIyP/RtehFRESClK5FLyIiUsSo4EVEREKQCl5ERCQEqeBFRERCkApeREQkBKngRUREQpAKXkREJASp4EVEREKQCl5ERCQEhdSV7MxsH7DDj7OsCvzmx/kVVfoe/UPfo3/oe/QPfY/+kd/vsbZzrlpWI0Kq4P3NzOKzuwSg5Jy+R//Q9+gf+h79Q9+jfwTye9QmehERkRCkghcREQlBKvgzm+R1gBCh79E/9D36h75H/9D36B8B+x61D15ERCQEaQ1eREQkBKngs2BmV5vZZjPbamZDvc4TLMzsAjNbbGYbzWyDmT3oG17ZzD43sx98/1byOmswMLNwM1trZrN9ryPN7Gvf73KqmZXwOmNhZ2YVzewjM9tkZglmdrl+j7lnZg/5/pv+3sz+Y2al9Hs8OzObbGa/mtn3mYZl+fuzdBN83+d3ZnZZfpevgj+FmYUDrwBdgUbAADNr5G2qoHEcGOKcawTEAPf5vruhwELnXH1goe+1nN2DQEKm1/8CXnDOXQgkAnd4kiq4vAjMc841BJqS/n3q95gLZlYDeACIcs5dAoQD/dHvMSf+DVx9yrDsfn9dgfq+x93Aa/lduAr+dK2Arc657c65o8B/gV4eZwoKzrndzrlvfM+TSP+faQ3Sv7+3fZO9DfT2JmHwMLOaQHfgTd9rAzoCH/km0fd4FmZWAbgSeAvAOXfUOXcQ/R7zohgQYWbFgNLAbvR7PCvn3JfAgVMGZ/f76wW849J9BVQ0s+r5Wb4K/nQ1gJ8zvd7pGya5YGZ1gObA18C5zrndvlF7gHM9ihVMxgOPAmm+11WAg865477X+l2eXSSwD5ji29XxppmVQb/HXHHO/QKMA/5HerEfAtag32NeZff783v3qODF78ysLPAxEOucO5x5nEs/bUOnbpyBmfUAfnXOrfE6S5ArBlwGvOacaw78zimb4/V7PDvfPuJepP/BdD5QhtM3O0seBPr3p4I/3S/ABZle1/QNkxwws+Kkl/v7zrlPfIP3Zmxq8v37q1f5gkQboKeZ/UT6LqKOpO9LrujbRAr6XebETmCnc+5r3+uPSC98/R5zpzPwo3Nun3PuGPAJ6b9R/R7zJrvfn9+7RwV/utVAfd8RoiVIP5gkzuNMQcG3n/gtIME593ymUXHALb7ntwAzCzpbMHHODXPO1XTO1SH997fIOXcDsBjo55tM3+NZOOf2AD+bWQPfoE7ARvR7zK3/ATFmVtr333jG96jfY95k9/uLA272HU0fAxzKtCk/T3ShmyyYWTfS94GGA5Odc//0OFJQMLO2wFJgPf+37/hx0vfDTwNqkX63v+udc6ceeCJZMLP2wN+ccz3MrC7pa/SVgbXAjc65FC/zFXZm1oz0AxVLANuB20hfsdHvMRfM7Cngr6SfKbMWuJP0/cP6PZ6Bmf0HaE/6HeP2Ak8CM8ji9+f74+ll0nd//AHc5pyLz9fyVfAiIiKhR5voRUREQpAKXkREJASp4EVEREKQCl5ERCQEqeBFRERCkApepIgzs1Qz+zbTw283XzGzOpnvpCUiBafY2ScRkRD3p3OumdchRMS/tAYvIlkys5/M7FkzW29mq8zsQt/wOma2yHfP6oVmVss3/Fwzm25m63yP1r5ZhZvZG777ic83swjf9A+Y2UbffP7r0ccUCVkqeBGJOGUT/V8zjTvknLuU9CtsjfcNewl42znXBHgfmOAbPgH4wjnXlPRrvm/wDa8PvOKcawwcBPr6hg8Fmvvmc0+gPpxIUaUr2YkUcWaW7Jwrm8Xwn4COzrntvpsI7XHOVTGz34DqzrljvuG7nXNVzWwfUDPz5Up9tw3+3DlX3/f6MaC4c260mc0Dkkm/dOcM51xygD+qSJGiNXgROROXzfPcyHx98lT+79if7sArpK/tr850ZzIR8QMVvIicyV8z/bvS93wF6Xe5A7iB9BsMASwE7gUws3Azq5DdTM0sDLjAObcYeAyoAJy2FUFE8k5/MYtIhJl9m+n1POdcxqlylczsO9LXwgf4ht0PTDGzR4B9pN+hDeBBYJKZ3UH6mvq9QHa3uwwH3vP9EWDABOfcQb99IhHRPngRyZpvH3yUc+43r7OISO5pE72IiEgI0hq8iIhICNIavIiISAhSwYuIiIQgFbyIiEgIUsGLiIiEIBW8iIhICFLBi4iIhKD/DwjQUiOjlSZBAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "cellView": "both",
        "id": "37tBXd11cdsD",
        "colab": {}
      },
      "source": [
        "#Guardar el modelo\n",
        "#@markdown optimizer_epochs_samples_features_accuracy_version\n",
        "\n",
        "nombre = \"LSTM_SGD_100e_650m_2f_1_V1\" #@param {type:\"string\"}\n",
        "model.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.h5')\n",
        "np.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.npy', le.classes_)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "ef4Bv3qP3Ge1",
        "colab": {}
      },
      "source": [
        "# Cargar modelo\n",
        "model = tf.keras.models.load_model('Models/model_LSTM_SGD_100e_650m_2f_1_V1.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "r4HsXbVl3GfJ",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "29aa0525-0992-42ea-c0ff-9d9666b05af7"
      },
      "source": [
        "#@title Predicción\n",
        "X_test = x_test.reshape(x_test.shape[0],x_test.shape[1],1)\n",
        "y_predict = model.predict_classes(X_test)\n",
        "y_test_single = np.argmax(y_test, axis=1)\n",
        "conf_matrix = confusion_matrix(y_test_single, y_predict)\n",
        "print(conf_matrix)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[ 35   4  12   5  79   6]\n",
            " [  4  77  27  41  36   3]\n",
            " [  6  24  48  16  68   6]\n",
            " [  4  27  14  70  34  13]\n",
            " [ 81  22  51  23 286  25]\n",
            " [  8   2   3  13  27 100]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "aAQo9hO23GfP",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "4cfe7051-4cbd-4501-9dfe-1ae4e66c6666"
      },
      "source": [
        "text = [\"ON\", \"OFF\", \"YES\", \"NO\", \"STOP\", \"UNKNOWN\"]\n",
        "#Vector para gráficas\n",
        "LSTM100=[]\n",
        "for i in range(len(text)):\n",
        "  acierto_promedio = conf_matrix[i][i]/(np.sum(conf_matrix[i], axis=0))\n",
        "  LSTM100.append(np.round(acierto_promedio, 2))\n",
        "  print(f\"Aciertos clase {text[i]}: {np.round(acierto_promedio*100, 2)}%\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Aciertos clase ON: 24.82%\n",
            "Aciertos clase OFF: 40.96%\n",
            "Aciertos clase YES: 28.57%\n",
            "Aciertos clase NO: 43.21%\n",
            "Aciertos clase STOP: 58.61%\n",
            "Aciertos clase UNKNOWN: 65.36%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "UguEnVHVKm33"
      },
      "source": [
        "### Resultados con LSTM"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MkvWXnzBd03v",
        "colab_type": "code",
        "cellView": "both",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "1e630c35-7d65-40a0-a2bb-bb9851ceb7b8"
      },
      "source": [
        "barWidth=0.2\n",
        "plt.bar(np.arange(len(LSTM50)), LSTM50, color=\"#6A5ACD\", width=barWidth, label=\"50 Epochs\");\n",
        "plt.bar([x + barWidth for x in np.arange(len(LSTM50))], LSTM100, color=\"#00BFFF\", width=barWidth, label=\"100 Epochs\");\n",
        "plt.xlabel(\"Clases\");\n",
        "plt.ylabel(\"Aciertos en % \");\n",
        "plt.xticks([r + barWidth for r in range(len(text))], text);\n",
        "plt.title(\"Resultados de la implementación con modelo LSTM\");\n",
        "plt.legend();\n",
        "plt.show();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RmpBNtqgQJOr",
        "colab_type": "text"
      },
      "source": [
        "##Entrenamiento Modelo GRU"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "q_8zNXlNQQjr",
        "colab_type": "code",
        "cellView": "both",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 353
        },
        "outputId": "10ab2bd8-ff08-4b9c-c3df-f6f546ae2f14"
      },
      "source": [
        "tf.keras.backend.clear_session()\n",
        "\n",
        "model = tf.keras.models.Sequential(name=\"Modelo_GRU\")\n",
        "# model.add(tf.keras.layers.Embedding(4, 300, input_length=40))\n",
        "model.add(tf.keras.layers.Input((92,1)))\n",
        "model.add(tf.keras.layers.Bidirectional(tf.keras.layers.GRU(units=512,\n",
        "                                                          return_sequences=True,\n",
        "                                                          activation='tanh',\n",
        "                                                          name='GRU_layer1')))\n",
        "model.add(tf.keras.layers.Bidirectional(tf.keras.layers.GRU(units=512,\n",
        "                                                          return_sequences=True,\n",
        "                                                          activation='tanh',\n",
        "                                                          name='GRU_layer2')))\n",
        "model.add(tf.keras.layers.Bidirectional(tf.keras.layers.GRU(units=256,\n",
        "                                                          return_sequences=False,\n",
        "                                                          activation='tanh',\n",
        "                                                          name='GRU_layer3')))\n",
        "model.add(tf.keras.layers.Dense(units=256, \n",
        "                                   activation='relu',\n",
        "                                   name='dense1') )\n",
        "model.add(tf.keras.layers.Dense(units=128, \n",
        "                                   activation='relu',\n",
        "                                   name='dense2') )\n",
        "model.add(tf.keras.layers.Dense(units = num_labels,\n",
        "                                  activation='softmax',\n",
        "                                  name='ouput_layer'))\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"Modelo_GRU\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "bidirectional (Bidirectional (None, 92, 1024)          1582080   \n",
            "_________________________________________________________________\n",
            "bidirectional_1 (Bidirection (None, 92, 1024)          4724736   \n",
            "_________________________________________________________________\n",
            "bidirectional_2 (Bidirection (None, 512)               1969152   \n",
            "_________________________________________________________________\n",
            "dense1 (Dense)               (None, 256)               131328    \n",
            "_________________________________________________________________\n",
            "dense2 (Dense)               (None, 128)               32896     \n",
            "_________________________________________________________________\n",
            "ouput_layer (Dense)          (None, 6)                 774       \n",
            "=================================================================\n",
            "Total params: 8,440,966\n",
            "Trainable params: 8,440,966\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "ykMZshnpeOGx"
      },
      "source": [
        "### 50 Epochs"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "cellView": "both",
        "id": "H0IvgElLeOG2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "0d6d4729-a229-46f5-8bf5-31aea2bc5f55"
      },
      "source": [
        "#@title Entrenamiento\n",
        "epochs = 50 #@param {type:\"slider\", min:50, max:5000, step:25}\n",
        "\n",
        "#optimizer = tf.keras.optimizers.SGD(lr=1e-2, momentum=0.9)\n",
        "optimizer = tf.keras.optimizers.Adam(lr=1e-4)\n",
        "#optimizer = tf.keras.optimizers.Adadelta(lr=1e-2)\n",
        "\n",
        "# optimizer = 'adam' #@param [\"adam\", \"SGD\", \"adadelta\"]\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              metrics=['accuracy'],\n",
        "              optimizer=optimizer)\n",
        "X = x_train.reshape(x_train.shape[0],x_train.shape[1],1)\n",
        "# print(X.shape, y_train.shape)\n",
        "history = model.fit(X, y_train,epochs=epochs,verbose=1)\n",
        "\n",
        "acc = history.history['accuracy']\n",
        "loss = history.history['loss']\n",
        "\n",
        "epochs=range(len(loss)) # Get number of epochs\n",
        "\n",
        "#------------------------------------------------\n",
        "# Plot acc and Loss\n",
        "#------------------------------------------------\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.plot(epochs, acc, 'red', label='acc')\n",
        "plt.title('Acc')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.show();\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.title('Loss')\n",
        "plt.plot(epochs, loss, 'black', label='loss')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Error\")\n",
        "plt.show();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.5968 - accuracy: 0.3887\n",
            "Epoch 2/50\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.4252 - accuracy: 0.4479\n",
            "Epoch 3/50\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.3523 - accuracy: 0.4774\n",
            "Epoch 4/50\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.3145 - accuracy: 0.4869\n",
            "Epoch 5/50\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.2944 - accuracy: 0.4990\n",
            "Epoch 6/50\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.2735 - accuracy: 0.5059\n",
            "Epoch 7/50\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.2438 - accuracy: 0.5190\n",
            "Epoch 8/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 1.2225 - accuracy: 0.5195\n",
            "Epoch 9/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 1.2027 - accuracy: 0.5269\n",
            "Epoch 10/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 1.1919 - accuracy: 0.5305\n",
            "Epoch 11/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 1.1707 - accuracy: 0.5344\n",
            "Epoch 12/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 1.1489 - accuracy: 0.5451\n",
            "Epoch 13/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 1.1214 - accuracy: 0.5533\n",
            "Epoch 14/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 1.0980 - accuracy: 0.5695\n",
            "Epoch 15/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 1.0725 - accuracy: 0.5738\n",
            "Epoch 16/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 1.0479 - accuracy: 0.5859\n",
            "Epoch 17/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 1.0247 - accuracy: 0.5951\n",
            "Epoch 18/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.9969 - accuracy: 0.6144\n",
            "Epoch 19/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.9704 - accuracy: 0.6215\n",
            "Epoch 20/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.9403 - accuracy: 0.6377\n",
            "Epoch 21/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.9028 - accuracy: 0.6562\n",
            "Epoch 22/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.8687 - accuracy: 0.6628\n",
            "Epoch 23/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.8315 - accuracy: 0.6833\n",
            "Epoch 24/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.8049 - accuracy: 0.6923\n",
            "Epoch 25/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.7567 - accuracy: 0.7141\n",
            "Epoch 26/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.7124 - accuracy: 0.7323\n",
            "Epoch 27/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.6625 - accuracy: 0.7592\n",
            "Epoch 28/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.6138 - accuracy: 0.7754\n",
            "Epoch 29/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.5675 - accuracy: 0.8008\n",
            "Epoch 30/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.5129 - accuracy: 0.8262\n",
            "Epoch 31/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.4669 - accuracy: 0.8441\n",
            "Epoch 32/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.4021 - accuracy: 0.8774\n",
            "Epoch 33/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.3440 - accuracy: 0.9021\n",
            "Epoch 34/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.2935 - accuracy: 0.9241\n",
            "Epoch 35/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.2486 - accuracy: 0.9403\n",
            "Epoch 36/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.2056 - accuracy: 0.9546\n",
            "Epoch 37/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.1628 - accuracy: 0.9721\n",
            "Epoch 38/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.1219 - accuracy: 0.9826\n",
            "Epoch 39/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.0880 - accuracy: 0.9931\n",
            "Epoch 40/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.0657 - accuracy: 0.9967\n",
            "Epoch 41/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.0509 - accuracy: 0.9985\n",
            "Epoch 42/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.0345 - accuracy: 0.9997\n",
            "Epoch 43/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.0239 - accuracy: 1.0000\n",
            "Epoch 44/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.0176 - accuracy: 1.0000\n",
            "Epoch 45/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.0143 - accuracy: 1.0000\n",
            "Epoch 46/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.0113 - accuracy: 1.0000\n",
            "Epoch 47/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.0090 - accuracy: 1.0000\n",
            "Epoch 48/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.0075 - accuracy: 1.0000\n",
            "Epoch 49/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.0062 - accuracy: 1.0000\n",
            "Epoch 50/50\n",
            "122/122 [==============================] - 9s 76ms/step - loss: 0.0053 - accuracy: 1.0000\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "cellView": "both",
        "id": "76MxT8KeeOG_",
        "colab": {}
      },
      "source": [
        "#Guardar el modelo\n",
        "#@markdown optimizer_epochs_samples_features_accuracy_version\n",
        "\n",
        "nombre = \"GRU_SGD_50e_650m_2f_1_V1\" #@param {type:\"string\"}\n",
        "model.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.h5')\n",
        "np.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.npy', le.classes_)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "4a53Kvci64Io",
        "colab": {}
      },
      "source": [
        "# Cargar modelo\n",
        "model = tf.keras.models.load_model('Models/model_GRU_SGD_50e_650m_2f_1_V1.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "yMfihS4T64JD",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "17ded157-a7e0-49e3-b38a-54e774156350"
      },
      "source": [
        "#@title Predicción\n",
        "X_test = x_test.reshape(x_test.shape[0],x_test.shape[1],1)\n",
        "y_predict = model.predict_classes(X_test)\n",
        "y_test_single = np.argmax(y_test, axis=1)\n",
        "conf_matrix = confusion_matrix(y_test_single, y_predict)\n",
        "print(conf_matrix)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[ 64   3  13   5  52   4]\n",
            " [  3 111  23  20  29   2]\n",
            " [  9  16  82   5  55   1]\n",
            " [  6  26   6  89  28   7]\n",
            " [ 55  20  39  16 340  18]\n",
            " [  3   6   0   6  18 120]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "k5Bd6AAB64JI",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "d2f8dda5-3833-43f5-a1e4-d9fc67f620ce"
      },
      "source": [
        "text = [\"ON\", \"OFF\", \"YES\", \"NO\", \"STOP\", \"UNKNOWN\"]\n",
        "#Vector para gráficas\n",
        "GRU50=[]\n",
        "for i in range(len(text)):\n",
        "  acierto_promedio = conf_matrix[i][i]/(np.sum(conf_matrix[i], axis=0))\n",
        "  GRU50.append(np.round(acierto_promedio, 2))\n",
        "  print(f\"Aciertos clase {text[i]}: {np.round(acierto_promedio*100, 2)}%\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Aciertos clase ON: 45.39%\n",
            "Aciertos clase OFF: 59.04%\n",
            "Aciertos clase YES: 48.81%\n",
            "Aciertos clase NO: 54.94%\n",
            "Aciertos clase STOP: 69.67%\n",
            "Aciertos clase UNKNOWN: 78.43%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BLTJbs7Wd-Ps",
        "colab_type": "text"
      },
      "source": [
        "### 100 Epochs"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zQ5wIVAkQP9M",
        "colab_type": "code",
        "cellView": "both",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "f41928dd-3eeb-44a9-d512-ae5b335097d6"
      },
      "source": [
        "#@title Entrenamiento\n",
        "epochs = 100 #@param {type:\"slider\", min:50, max:5000, step:25}\n",
        "\n",
        "#optimizer = tf.keras.optimizers.SGD(lr=1e-2, momentum=0.9)\n",
        "optimizer = tf.keras.optimizers.Adam(lr=1e-4)\n",
        "#optimizer = tf.keras.optimizers.Adadelta(lr=1e-2)\n",
        "\n",
        "# optimizer = 'adam' #@param [\"adam\", \"SGD\", \"adadelta\"]\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              metrics=['accuracy'],\n",
        "              optimizer=optimizer)\n",
        "X = x_train.reshape(x_train.shape[0],x_train.shape[1],1)\n",
        "# print(X.shape, y_train.shape)\n",
        "history = model.fit(X, y_train,epochs=epochs,verbose=1)\n",
        "\n",
        "acc = history.history['accuracy']\n",
        "loss = history.history['loss']\n",
        "\n",
        "epochs=range(len(loss)) # Get number of epochs\n",
        "\n",
        "#------------------------------------------------\n",
        "# Plot acc and Loss\n",
        "#------------------------------------------------\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.plot(epochs, acc, 'red', label='acc')\n",
        "plt.title('Acc')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.show();\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.title('Loss')\n",
        "plt.plot(epochs, loss, 'black', label='loss')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Error\")\n",
        "plt.show();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.5797 - accuracy: 0.3949\n",
            "Epoch 2/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.4171 - accuracy: 0.4541\n",
            "Epoch 3/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.3170 - accuracy: 0.4977\n",
            "Epoch 4/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.2791 - accuracy: 0.5026\n",
            "Epoch 5/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.2629 - accuracy: 0.5108\n",
            "Epoch 6/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.2443 - accuracy: 0.5172\n",
            "Epoch 7/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.2208 - accuracy: 0.5290\n",
            "Epoch 8/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.1951 - accuracy: 0.5403\n",
            "Epoch 9/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.1832 - accuracy: 0.5328\n",
            "Epoch 10/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.1554 - accuracy: 0.5433\n",
            "Epoch 11/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.1472 - accuracy: 0.5582\n",
            "Epoch 12/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.1112 - accuracy: 0.5679\n",
            "Epoch 13/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.1023 - accuracy: 0.5667\n",
            "Epoch 14/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.0775 - accuracy: 0.5810\n",
            "Epoch 15/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.0482 - accuracy: 0.5897\n",
            "Epoch 16/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.0298 - accuracy: 0.5954\n",
            "Epoch 17/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.0116 - accuracy: 0.6049\n",
            "Epoch 18/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.9855 - accuracy: 0.6195\n",
            "Epoch 19/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.9620 - accuracy: 0.6179\n",
            "Epoch 20/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.9249 - accuracy: 0.6405\n",
            "Epoch 21/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.8888 - accuracy: 0.6541\n",
            "Epoch 22/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.8652 - accuracy: 0.6600\n",
            "Epoch 23/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 0.8181 - accuracy: 0.6890\n",
            "Epoch 24/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.7851 - accuracy: 0.6959\n",
            "Epoch 25/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.7430 - accuracy: 0.7159\n",
            "Epoch 26/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.7068 - accuracy: 0.7395\n",
            "Epoch 27/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.6669 - accuracy: 0.7487\n",
            "Epoch 28/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.6285 - accuracy: 0.7662\n",
            "Epoch 29/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 0.5642 - accuracy: 0.8051\n",
            "Epoch 30/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.5213 - accuracy: 0.8187\n",
            "Epoch 31/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.4632 - accuracy: 0.8405\n",
            "Epoch 32/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.4192 - accuracy: 0.8651\n",
            "Epoch 33/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.3634 - accuracy: 0.8831\n",
            "Epoch 34/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.2984 - accuracy: 0.9218\n",
            "Epoch 35/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.2606 - accuracy: 0.9269\n",
            "Epoch 36/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.2109 - accuracy: 0.9515\n",
            "Epoch 37/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.1666 - accuracy: 0.9662\n",
            "Epoch 38/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 0.1309 - accuracy: 0.9769\n",
            "Epoch 39/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0917 - accuracy: 0.9905\n",
            "Epoch 40/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0634 - accuracy: 0.9977\n",
            "Epoch 41/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 0.0443 - accuracy: 0.9995\n",
            "Epoch 42/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0351 - accuracy: 0.9985\n",
            "Epoch 43/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0249 - accuracy: 0.9997\n",
            "Epoch 44/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0173 - accuracy: 1.0000\n",
            "Epoch 45/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0134 - accuracy: 1.0000\n",
            "Epoch 46/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0106 - accuracy: 1.0000\n",
            "Epoch 47/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 0.0086 - accuracy: 1.0000\n",
            "Epoch 48/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0072 - accuracy: 1.0000\n",
            "Epoch 49/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0060 - accuracy: 1.0000\n",
            "Epoch 50/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0051 - accuracy: 1.0000\n",
            "Epoch 51/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0044 - accuracy: 1.0000\n",
            "Epoch 52/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0038 - accuracy: 1.0000\n",
            "Epoch 53/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0033 - accuracy: 1.0000\n",
            "Epoch 54/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0028 - accuracy: 1.0000\n",
            "Epoch 55/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0025 - accuracy: 1.0000\n",
            "Epoch 56/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 0.0021 - accuracy: 1.0000\n",
            "Epoch 57/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 0.0019 - accuracy: 1.0000\n",
            "Epoch 58/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 0.0017 - accuracy: 1.0000\n",
            "Epoch 59/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 0.0015 - accuracy: 1.0000\n",
            "Epoch 60/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 0.0013 - accuracy: 1.0000\n",
            "Epoch 61/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 0.0012 - accuracy: 1.0000\n",
            "Epoch 62/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 0.0010 - accuracy: 1.0000\n",
            "Epoch 63/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 9.2337e-04 - accuracy: 1.0000\n",
            "Epoch 64/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 8.2267e-04 - accuracy: 1.0000\n",
            "Epoch 65/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 7.4619e-04 - accuracy: 1.0000\n",
            "Epoch 66/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 6.6656e-04 - accuracy: 1.0000\n",
            "Epoch 67/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 6.0367e-04 - accuracy: 1.0000\n",
            "Epoch 68/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 5.4233e-04 - accuracy: 1.0000\n",
            "Epoch 69/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 4.8976e-04 - accuracy: 1.0000\n",
            "Epoch 70/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 4.4086e-04 - accuracy: 1.0000\n",
            "Epoch 71/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 3.9936e-04 - accuracy: 1.0000\n",
            "Epoch 72/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 3.6289e-04 - accuracy: 1.0000\n",
            "Epoch 73/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 3.2832e-04 - accuracy: 1.0000\n",
            "Epoch 74/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 2.9893e-04 - accuracy: 1.0000\n",
            "Epoch 75/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 2.7123e-04 - accuracy: 1.0000\n",
            "Epoch 76/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 2.4662e-04 - accuracy: 1.0000\n",
            "Epoch 77/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 2.2530e-04 - accuracy: 1.0000\n",
            "Epoch 78/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 2.0467e-04 - accuracy: 1.0000\n",
            "Epoch 79/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.8766e-04 - accuracy: 1.0000\n",
            "Epoch 80/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.7031e-04 - accuracy: 1.0000\n",
            "Epoch 81/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.5499e-04 - accuracy: 1.0000\n",
            "Epoch 82/100\n",
            "122/122 [==============================] - 9s 74ms/step - loss: 1.4273e-04 - accuracy: 1.0000\n",
            "Epoch 83/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.2978e-04 - accuracy: 1.0000\n",
            "Epoch 84/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.1901e-04 - accuracy: 1.0000\n",
            "Epoch 85/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 1.0952e-04 - accuracy: 1.0000\n",
            "Epoch 86/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 9.9716e-05 - accuracy: 1.0000\n",
            "Epoch 87/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 9.1499e-05 - accuracy: 1.0000\n",
            "Epoch 88/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 8.4578e-05 - accuracy: 1.0000\n",
            "Epoch 89/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 7.7227e-05 - accuracy: 1.0000\n",
            "Epoch 90/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 7.0885e-05 - accuracy: 1.0000\n",
            "Epoch 91/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 6.4913e-05 - accuracy: 1.0000\n",
            "Epoch 92/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 5.9717e-05 - accuracy: 1.0000\n",
            "Epoch 93/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 5.4897e-05 - accuracy: 1.0000\n",
            "Epoch 94/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 5.0669e-05 - accuracy: 1.0000\n",
            "Epoch 95/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 4.6761e-05 - accuracy: 1.0000\n",
            "Epoch 96/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 4.3157e-05 - accuracy: 1.0000\n",
            "Epoch 97/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 3.9649e-05 - accuracy: 1.0000\n",
            "Epoch 98/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 3.6401e-05 - accuracy: 1.0000\n",
            "Epoch 99/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 3.3647e-05 - accuracy: 1.0000\n",
            "Epoch 100/100\n",
            "122/122 [==============================] - 9s 75ms/step - loss: 3.1058e-05 - accuracy: 1.0000\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nk2V_7JfQPYo",
        "colab_type": "code",
        "cellView": "both",
        "colab": {}
      },
      "source": [
        "#uardar modelo\n",
        "#@markdown optimizer_epochs_samples_features_accuracy_version\n",
        "\n",
        "nombre = \"GRU_SGD_100e_650m_2f_1_V1\" #@param {type:\"string\"}\n",
        "model.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.h5')\n",
        "np.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.npy', le.classes_)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "_yqAcFms79nJ",
        "colab": {}
      },
      "source": [
        "# Cargar modelo\n",
        "model = tf.keras.models.load_model('/content/drive/My Drive/IA2/DATA/Models/model_GRU_SGD_100e_650m_2f_1_V1.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "E0R8mZkc79nb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "32366a8d-df25-4c65-e3b5-fcb65beded87"
      },
      "source": [
        "#@title Predicción\n",
        "X_test = x_test.reshape(x_test.shape[0],x_test.shape[1],1)\n",
        "y_predict = model.predict_classes(X_test)\n",
        "y_test_single = np.argmax(y_test, axis=1)\n",
        "conf_matrix = confusion_matrix(y_test_single, y_predict)\n",
        "print(conf_matrix)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[ 47   0  20   4  65   5]\n",
            " [  5 104  20  29  28   2]\n",
            " [ 14  27  50  16  58   3]\n",
            " [  7  23  14  78  34   6]\n",
            " [ 79  31  57  42 263  16]\n",
            " [  8   3   2   9  26 105]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "FELJ-w7z79nk",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "d99dd3f5-88bc-476a-d15f-a677eeb93999"
      },
      "source": [
        "text = [\"ON\", \"OFF\", \"YES\", \"NO\", \"STOP\", \"UNKNOWN\"]\n",
        "#Vector para gráficas\n",
        "GRU100=[]\n",
        "for i in range(len(text)):\n",
        "  acierto_promedio = conf_matrix[i][i]/(np.sum(conf_matrix[i], axis=0))\n",
        "  GRU100.append(np.round(acierto_promedio, 2))\n",
        "  print(f\"Aciertos clase {text[i]}: {np.round(acierto_promedio*100, 2)}%\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Aciertos clase ON: 33.33%\n",
            "Aciertos clase OFF: 55.32%\n",
            "Aciertos clase YES: 29.76%\n",
            "Aciertos clase NO: 48.15%\n",
            "Aciertos clase STOP: 53.89%\n",
            "Aciertos clase UNKNOWN: 68.63%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "8RmBzmnAMl4q"
      },
      "source": [
        "### Resultados con GRU"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3DLy1C6AgsR-",
        "colab_type": "code",
        "cellView": "both",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "a6c6f7bb-9790-401a-cf38-a19da9c5aa89"
      },
      "source": [
        "barWidth=0.2\n",
        "plt.bar(np.arange(len(GRU50)), GRU50, color=\"#6A5ACD\", width=barWidth, label=\"50 Epochs\");\n",
        "plt.bar([x + barWidth for x in np.arange(len(GRU50))], GRU100, color=\"#00BFFF\", width=barWidth, label=\"100 Epochs\");\n",
        "plt.xlabel(\"Clases\");\n",
        "plt.ylabel(\"Aciertos en %\");\n",
        "plt.xticks([r + barWidth for r in range(len(text))], text);\n",
        "plt.title(\"Resultados de la implementación con modelo GRU\");\n",
        "plt.legend();\n",
        "plt.show();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2AaPVkzZ_UUU",
        "colab_type": "text"
      },
      "source": [
        "##Entrenamiento Modelo DNN"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2ydCFFzm_omG",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 527
        },
        "outputId": "19cdf0a9-166e-4be5-8360-dd7a697f74b8"
      },
      "source": [
        "K.clear_session()\n",
        "\n",
        "inputs = Input(shape=(92,1))\n",
        "\n",
        "#First Conv1D layer\n",
        "conv = Conv1D(26,13, padding='valid', activation='tanh', strides=1)(inputs)\n",
        "conv = MaxPooling1D(pool_size= 3)(conv)\n",
        "# conv = Dropout(0.3)(conv)\n",
        "\n",
        "#Second Conv1D layer\n",
        "conv = Conv1D(20, 11, padding='valid', activation='relu', strides=1)(conv)\n",
        "conv = MaxPooling1D(pool_size=3)(conv)\n",
        "# conv = Dropout(0.3)(conv)\n",
        "\n",
        "#Third Conv1D layer\n",
        "conv = Conv1D(10, 5, padding='valid', activation='relu', strides=1)(conv)\n",
        "# conv = MaxPooling1D(pool_size=0)(conv)\n",
        "# conv = Dropout(0.3)(conv)\n",
        "\n",
        "# #Fourth Conv1D layer\n",
        "# conv = Conv1D(5, 1, activation='relu')(conv)\n",
        "# # conv = MaxPooling1D(pool_size=3)(conv)\n",
        "# conv = Dropout(0.3)(conv)\n",
        "\n",
        "#Flatten layer\n",
        "conv = Flatten()(conv)\n",
        "\n",
        "#Dense Layer 1\n",
        "conv = Dense(256, activation='relu')(conv)\n",
        "# conv = Dropout(0.3)(conv)\n",
        "\n",
        "#Dense Layer 2\n",
        "conv = Dense(128, activation='relu')(conv)\n",
        "# conv = Dropout(0.3)(conv)\n",
        "\n",
        "#Dense Layer 2\n",
        "conv = Dense(64, activation='relu')(conv)\n",
        "\n",
        "outputs = Dense(num_labels, activation='softmax')(conv)\n",
        "\n",
        "model = Model(inputs, outputs)\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"functional_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_1 (InputLayer)         [(None, 92, 1)]           0         \n",
            "_________________________________________________________________\n",
            "conv1d (Conv1D)              (None, 80, 26)            364       \n",
            "_________________________________________________________________\n",
            "max_pooling1d (MaxPooling1D) (None, 26, 26)            0         \n",
            "_________________________________________________________________\n",
            "conv1d_1 (Conv1D)            (None, 16, 20)            5740      \n",
            "_________________________________________________________________\n",
            "max_pooling1d_1 (MaxPooling1 (None, 5, 20)             0         \n",
            "_________________________________________________________________\n",
            "conv1d_2 (Conv1D)            (None, 1, 10)             1010      \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 10)                0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 256)               2816      \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 128)               32896     \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 64)                8256      \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 6)                 390       \n",
            "=================================================================\n",
            "Total params: 51,472\n",
            "Trainable params: 51,472\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-7A_VhoCinFZ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "9371f400-e131-4ebf-ec60-227edefd376b"
      },
      "source": [
        "#@title Entrenamiento con Adam\n",
        "epochs = 200 #@param {type:\"slider\", min:50, max:5000, step:25}\n",
        "\n",
        "#optimizer = tf.keras.optimizers.SGD(lr=1e-2, momentum=0.9)\n",
        "optimizer = tf.keras.optimizers.Adam(lr=1e-4)\n",
        "#optimizer = tf.keras.optimizers.Adadelta(lr=1e-2)\n",
        "\n",
        "# optimizer = 'adam' #@param [\"adam\", \"SGD\", \"adadelta\"]\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              metrics=['accuracy'],\n",
        "              optimizer=optimizer)\n",
        "X = x_train.reshape(x_train.shape[0],x_train.shape[1],1)\n",
        "# print(X.shape, y_train.shape)\n",
        "history = model.fit(X, y_train,epochs=epochs,verbose=1)\n",
        "\n",
        "acc = history.history['accuracy']\n",
        "loss = history.history['loss']\n",
        "\n",
        "epochs=range(len(loss)) # Get number of epochs\n",
        "\n",
        "#------------------------------------------------\n",
        "# Plot acc and Loss\n",
        "#------------------------------------------------\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.plot(epochs, acc, 'red', label='acc')\n",
        "plt.title('Acc')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.show();\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.title('Loss')\n",
        "plt.plot(epochs, loss, 'black', label='loss')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Error\")\n",
        "plt.show();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.7173 - accuracy: 0.3449\n",
            "Epoch 2/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6572 - accuracy: 0.3733\n",
            "Epoch 3/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6466 - accuracy: 0.3733\n",
            "Epoch 4/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6360 - accuracy: 0.3733\n",
            "Epoch 5/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6236 - accuracy: 0.3733\n",
            "Epoch 6/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6069 - accuracy: 0.3733\n",
            "Epoch 7/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.5869 - accuracy: 0.3733\n",
            "Epoch 8/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.5614 - accuracy: 0.3792\n",
            "Epoch 9/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.5365 - accuracy: 0.3928\n",
            "Epoch 10/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.5091 - accuracy: 0.4090\n",
            "Epoch 11/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.4900 - accuracy: 0.4113\n",
            "Epoch 12/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.4686 - accuracy: 0.4126\n",
            "Epoch 13/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.4508 - accuracy: 0.4208\n",
            "Epoch 14/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.4409 - accuracy: 0.4259\n",
            "Epoch 15/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.4214 - accuracy: 0.4369\n",
            "Epoch 16/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.4069 - accuracy: 0.4354\n",
            "Epoch 17/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3926 - accuracy: 0.4433\n",
            "Epoch 18/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3791 - accuracy: 0.4449\n",
            "Epoch 19/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3688 - accuracy: 0.4490\n",
            "Epoch 20/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3572 - accuracy: 0.4544\n",
            "Epoch 21/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3437 - accuracy: 0.4638\n",
            "Epoch 22/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3282 - accuracy: 0.4654\n",
            "Epoch 23/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3205 - accuracy: 0.4733\n",
            "Epoch 24/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3115 - accuracy: 0.4731\n",
            "Epoch 25/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2980 - accuracy: 0.4787\n",
            "Epoch 26/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2855 - accuracy: 0.4849\n",
            "Epoch 27/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2815 - accuracy: 0.4877\n",
            "Epoch 28/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2712 - accuracy: 0.4859\n",
            "Epoch 29/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2576 - accuracy: 0.4990\n",
            "Epoch 30/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2508 - accuracy: 0.4959\n",
            "Epoch 31/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2383 - accuracy: 0.4982\n",
            "Epoch 32/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2326 - accuracy: 0.5033\n",
            "Epoch 33/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2211 - accuracy: 0.5087\n",
            "Epoch 34/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2098 - accuracy: 0.5167\n",
            "Epoch 35/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2053 - accuracy: 0.5154\n",
            "Epoch 36/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1953 - accuracy: 0.5126\n",
            "Epoch 37/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1879 - accuracy: 0.5185\n",
            "Epoch 38/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1802 - accuracy: 0.5305\n",
            "Epoch 39/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1707 - accuracy: 0.5285\n",
            "Epoch 40/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1628 - accuracy: 0.5344\n",
            "Epoch 41/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1545 - accuracy: 0.5367\n",
            "Epoch 42/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1430 - accuracy: 0.5405\n",
            "Epoch 43/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1373 - accuracy: 0.5451\n",
            "Epoch 44/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1305 - accuracy: 0.5467\n",
            "Epoch 45/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1182 - accuracy: 0.5505\n",
            "Epoch 46/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1112 - accuracy: 0.5518\n",
            "Epoch 47/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1022 - accuracy: 0.5603\n",
            "Epoch 48/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0987 - accuracy: 0.5654\n",
            "Epoch 49/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0890 - accuracy: 0.5636\n",
            "Epoch 50/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0880 - accuracy: 0.5679\n",
            "Epoch 51/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0774 - accuracy: 0.5690\n",
            "Epoch 52/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0684 - accuracy: 0.5749\n",
            "Epoch 53/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0617 - accuracy: 0.5797\n",
            "Epoch 54/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0544 - accuracy: 0.5800\n",
            "Epoch 55/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0497 - accuracy: 0.5833\n",
            "Epoch 56/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0414 - accuracy: 0.5833\n",
            "Epoch 57/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0360 - accuracy: 0.5926\n",
            "Epoch 58/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0283 - accuracy: 0.5936\n",
            "Epoch 59/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0222 - accuracy: 0.5938\n",
            "Epoch 60/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0119 - accuracy: 0.6000\n",
            "Epoch 61/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0063 - accuracy: 0.6069\n",
            "Epoch 62/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0002 - accuracy: 0.6038\n",
            "Epoch 63/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9928 - accuracy: 0.6092\n",
            "Epoch 64/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9833 - accuracy: 0.6133\n",
            "Epoch 65/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9811 - accuracy: 0.6095\n",
            "Epoch 66/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9713 - accuracy: 0.6218\n",
            "Epoch 67/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9660 - accuracy: 0.6151\n",
            "Epoch 68/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9624 - accuracy: 0.6221\n",
            "Epoch 69/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9555 - accuracy: 0.6251\n",
            "Epoch 70/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9460 - accuracy: 0.6336\n",
            "Epoch 71/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9409 - accuracy: 0.6305\n",
            "Epoch 72/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9389 - accuracy: 0.6241\n",
            "Epoch 73/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9252 - accuracy: 0.6400\n",
            "Epoch 74/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9214 - accuracy: 0.6444\n",
            "Epoch 75/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9131 - accuracy: 0.6459\n",
            "Epoch 76/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9074 - accuracy: 0.6408\n",
            "Epoch 77/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9021 - accuracy: 0.6505\n",
            "Epoch 78/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8944 - accuracy: 0.6446\n",
            "Epoch 79/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8892 - accuracy: 0.6487\n",
            "Epoch 80/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8862 - accuracy: 0.6472\n",
            "Epoch 81/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8809 - accuracy: 0.6546\n",
            "Epoch 82/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8757 - accuracy: 0.6549\n",
            "Epoch 83/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8687 - accuracy: 0.6646\n",
            "Epoch 84/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8602 - accuracy: 0.6600\n",
            "Epoch 85/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8547 - accuracy: 0.6667\n",
            "Epoch 86/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8483 - accuracy: 0.6674\n",
            "Epoch 87/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8436 - accuracy: 0.6733\n",
            "Epoch 88/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8366 - accuracy: 0.6762\n",
            "Epoch 89/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8381 - accuracy: 0.6754\n",
            "Epoch 90/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8256 - accuracy: 0.6844\n",
            "Epoch 91/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8228 - accuracy: 0.6764\n",
            "Epoch 92/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8158 - accuracy: 0.6877\n",
            "Epoch 93/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8108 - accuracy: 0.6874\n",
            "Epoch 94/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8031 - accuracy: 0.6908\n",
            "Epoch 95/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7967 - accuracy: 0.6908\n",
            "Epoch 96/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7985 - accuracy: 0.6913\n",
            "Epoch 97/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7910 - accuracy: 0.6890\n",
            "Epoch 98/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7816 - accuracy: 0.6979\n",
            "Epoch 99/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7745 - accuracy: 0.7087\n",
            "Epoch 100/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7717 - accuracy: 0.7100\n",
            "Epoch 101/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7658 - accuracy: 0.7026\n",
            "Epoch 102/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7650 - accuracy: 0.7077\n",
            "Epoch 103/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7584 - accuracy: 0.7105\n",
            "Epoch 104/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7547 - accuracy: 0.7077\n",
            "Epoch 105/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7495 - accuracy: 0.7128\n",
            "Epoch 106/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7473 - accuracy: 0.7177\n",
            "Epoch 107/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7342 - accuracy: 0.7213\n",
            "Epoch 108/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7327 - accuracy: 0.7192\n",
            "Epoch 109/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7242 - accuracy: 0.7244\n",
            "Epoch 110/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7210 - accuracy: 0.7210\n",
            "Epoch 111/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7165 - accuracy: 0.7267\n",
            "Epoch 112/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7139 - accuracy: 0.7323\n",
            "Epoch 113/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7119 - accuracy: 0.7285\n",
            "Epoch 114/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7029 - accuracy: 0.7292\n",
            "Epoch 115/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6947 - accuracy: 0.7377\n",
            "Epoch 116/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6881 - accuracy: 0.7359\n",
            "Epoch 117/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6875 - accuracy: 0.7341\n",
            "Epoch 118/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6844 - accuracy: 0.7400\n",
            "Epoch 119/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6816 - accuracy: 0.7403\n",
            "Epoch 120/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6762 - accuracy: 0.7415\n",
            "Epoch 121/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6687 - accuracy: 0.7515\n",
            "Epoch 122/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6649 - accuracy: 0.7490\n",
            "Epoch 123/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6628 - accuracy: 0.7459\n",
            "Epoch 124/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6493 - accuracy: 0.7549\n",
            "Epoch 125/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6594 - accuracy: 0.7400\n",
            "Epoch 126/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6448 - accuracy: 0.7549\n",
            "Epoch 127/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6418 - accuracy: 0.7572\n",
            "Epoch 128/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6347 - accuracy: 0.7633\n",
            "Epoch 129/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6300 - accuracy: 0.7674\n",
            "Epoch 130/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6248 - accuracy: 0.7646\n",
            "Epoch 131/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6197 - accuracy: 0.7631\n",
            "Epoch 132/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6171 - accuracy: 0.7690\n",
            "Epoch 133/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6105 - accuracy: 0.7710\n",
            "Epoch 134/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6143 - accuracy: 0.7646\n",
            "Epoch 135/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6032 - accuracy: 0.7772\n",
            "Epoch 136/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6026 - accuracy: 0.7764\n",
            "Epoch 137/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6034 - accuracy: 0.7738\n",
            "Epoch 138/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5923 - accuracy: 0.7726\n",
            "Epoch 139/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5883 - accuracy: 0.7738\n",
            "Epoch 140/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5835 - accuracy: 0.7803\n",
            "Epoch 141/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5803 - accuracy: 0.7879\n",
            "Epoch 142/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5734 - accuracy: 0.7849\n",
            "Epoch 143/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5714 - accuracy: 0.7890\n",
            "Epoch 144/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5678 - accuracy: 0.7831\n",
            "Epoch 145/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5626 - accuracy: 0.7854\n",
            "Epoch 146/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5617 - accuracy: 0.7872\n",
            "Epoch 147/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5485 - accuracy: 0.7941\n",
            "Epoch 148/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5498 - accuracy: 0.7962\n",
            "Epoch 149/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5435 - accuracy: 0.7985\n",
            "Epoch 150/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5512 - accuracy: 0.7926\n",
            "Epoch 151/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5440 - accuracy: 0.7951\n",
            "Epoch 152/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5377 - accuracy: 0.8049\n",
            "Epoch 153/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5306 - accuracy: 0.8028\n",
            "Epoch 154/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5266 - accuracy: 0.8077\n",
            "Epoch 155/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5235 - accuracy: 0.8036\n",
            "Epoch 156/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5215 - accuracy: 0.8105\n",
            "Epoch 157/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5099 - accuracy: 0.8108\n",
            "Epoch 158/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5057 - accuracy: 0.8167\n",
            "Epoch 159/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5092 - accuracy: 0.8123\n",
            "Epoch 160/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5091 - accuracy: 0.8082\n",
            "Epoch 161/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4970 - accuracy: 0.8205\n",
            "Epoch 162/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4912 - accuracy: 0.8182\n",
            "Epoch 163/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4949 - accuracy: 0.8162\n",
            "Epoch 164/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4912 - accuracy: 0.8210\n",
            "Epoch 165/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4924 - accuracy: 0.8164\n",
            "Epoch 166/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4892 - accuracy: 0.8169\n",
            "Epoch 167/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4712 - accuracy: 0.8277\n",
            "Epoch 168/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4763 - accuracy: 0.8246\n",
            "Epoch 169/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4704 - accuracy: 0.8249\n",
            "Epoch 170/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4658 - accuracy: 0.8310\n",
            "Epoch 171/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4639 - accuracy: 0.8313\n",
            "Epoch 172/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4554 - accuracy: 0.8290\n",
            "Epoch 173/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4553 - accuracy: 0.8351\n",
            "Epoch 174/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4567 - accuracy: 0.8364\n",
            "Epoch 175/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4450 - accuracy: 0.8372\n",
            "Epoch 176/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4495 - accuracy: 0.8359\n",
            "Epoch 177/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4407 - accuracy: 0.8438\n",
            "Epoch 178/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4482 - accuracy: 0.8308\n",
            "Epoch 179/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4352 - accuracy: 0.8441\n",
            "Epoch 180/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4297 - accuracy: 0.8464\n",
            "Epoch 181/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4289 - accuracy: 0.8467\n",
            "Epoch 182/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4315 - accuracy: 0.8385\n",
            "Epoch 183/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4251 - accuracy: 0.8451\n",
            "Epoch 184/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4151 - accuracy: 0.8513\n",
            "Epoch 185/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4131 - accuracy: 0.8508\n",
            "Epoch 186/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4057 - accuracy: 0.8559\n",
            "Epoch 187/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4130 - accuracy: 0.8487\n",
            "Epoch 188/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4020 - accuracy: 0.8546\n",
            "Epoch 189/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.3987 - accuracy: 0.8572\n",
            "Epoch 190/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.3909 - accuracy: 0.8623\n",
            "Epoch 191/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.3948 - accuracy: 0.8631\n",
            "Epoch 192/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.3934 - accuracy: 0.8559\n",
            "Epoch 193/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.3943 - accuracy: 0.8585\n",
            "Epoch 194/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.4003 - accuracy: 0.8546\n",
            "Epoch 195/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.3843 - accuracy: 0.8654\n",
            "Epoch 196/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.3758 - accuracy: 0.8618\n",
            "Epoch 197/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.3725 - accuracy: 0.8646\n",
            "Epoch 198/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.3684 - accuracy: 0.8682\n",
            "Epoch 199/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.3641 - accuracy: 0.8705\n",
            "Epoch 200/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.3675 - accuracy: 0.8703\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "cellView": "both",
        "id": "-YvsyuGnVoul",
        "colab": {}
      },
      "source": [
        "#Guardar\n",
        "#@markdown optimizer_epochs_samples_features_accuracy_version\n",
        "\n",
        "nombre = \"DNN_Adamax_200e_650m_2f_0.87_V1\" #@param {type:\"string\"}\n",
        "model.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.h5')\n",
        "np.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.npy', le.classes_)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "ebFYalbMVout",
        "colab": {}
      },
      "source": [
        "# Cargar modelo\n",
        "model = tf.keras.models.load_model('/content/drive/My Drive/IA2/DATA/Models/model_DNN_Adamax_200e_650m_2f_0.87_V1.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "MVOWY_xUVouw",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "8e0541b3-6a1c-4cf7-a560-387567802b24"
      },
      "source": [
        "#@title Predicción\n",
        "X_test = x_test.reshape(x_test.shape[0],x_test.shape[1],1)\n",
        "y_predict = np.argmax(model.predict(X_test), axis=1)\n",
        "y_test_single = np.argmax(y_test, axis=1)\n",
        "conf_matrix = confusion_matrix(y_test_single, y_predict)\n",
        "print(conf_matrix)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[ 40   8   9   6  75   3]\n",
            " [  3  76  18  34  47  10]\n",
            " [ 11  15  51   8  76   7]\n",
            " [  5  19   4  86  29  19]\n",
            " [ 53  36  29  27 317  26]\n",
            " [  3   6   3  21  26  94]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "gaW-rvYsVouz",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "eac89384-d3df-4d23-e400-f84b194b8fa1"
      },
      "source": [
        "text = [\"ON\", \"OFF\", \"YES\", \"NO\", \"STOP\", \"UNKNOWN\"]\n",
        "#Vector para gráficas\n",
        "DNN_ADAM=[]\n",
        "for i in range(len(text)):\n",
        "  acierto_promedio = conf_matrix[i][i]/(np.sum(conf_matrix[i], axis=0))\n",
        "  DNN_ADAM.append(np.round(acierto_promedio, 2))\n",
        "  print(f\"Aciertos clase {text[i]}: {np.round(acierto_promedio*100, 2)}%\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Aciertos clase ON: 28.37%\n",
            "Aciertos clase OFF: 40.43%\n",
            "Aciertos clase YES: 30.36%\n",
            "Aciertos clase NO: 53.09%\n",
            "Aciertos clase STOP: 64.96%\n",
            "Aciertos clase UNKNOWN: 61.44%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2q24-DSzWJn8",
        "colab_type": "text"
      },
      "source": [
        "---"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "PyIe68pSoSQO",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "d9e1f115-7bf8-4c6e-c9ba-91855b551182"
      },
      "source": [
        "#@title Entrenamiento con SGD\n",
        "epochs = 200 #@param {type:\"slider\", min:50, max:5000, step:25}\n",
        "\n",
        "optimizer = tf.keras.optimizers.SGD(lr=1e-2, momentum=0.9)\n",
        "# optimizer = tf.keras.optimizers.Adam(lr=1e-4)\n",
        "#optimizer = tf.keras.optimizers.Adadelta(lr=1e-2)\n",
        "\n",
        "# optimizer = 'adam' #@param [\"adam\", \"SGD\", \"adadelta\"]\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              metrics=['accuracy'],\n",
        "              optimizer=optimizer)\n",
        "X = x_train.reshape(x_train.shape[0],x_train.shape[1],1)\n",
        "# print(X.shape, y_train.shape)\n",
        "history = model.fit(X, y_train,epochs=epochs,verbose=1)\n",
        "\n",
        "acc = history.history['accuracy']\n",
        "loss = history.history['loss']\n",
        "\n",
        "epochs=range(len(loss)) # Get number of epochs\n",
        "\n",
        "#------------------------------------------------\n",
        "# Plot acc and Loss\n",
        "#------------------------------------------------\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.plot(epochs, acc, 'red', label='acc')\n",
        "plt.title('Acc')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.show();\n",
        "plt.figure(figsize=(8, 6));\n",
        "plt.title('Loss')\n",
        "plt.plot(epochs, loss, 'black', label='loss')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Error\")\n",
        "plt.show();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.9703 - accuracy: 0.3595\n",
            "Epoch 2/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6715 - accuracy: 0.3733\n",
            "Epoch 3/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6718 - accuracy: 0.3733\n",
            "Epoch 4/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6720 - accuracy: 0.3733\n",
            "Epoch 5/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6712 - accuracy: 0.3736\n",
            "Epoch 6/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6705 - accuracy: 0.3738\n",
            "Epoch 7/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6689 - accuracy: 0.3738\n",
            "Epoch 8/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6675 - accuracy: 0.3736\n",
            "Epoch 9/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6604 - accuracy: 0.3733\n",
            "Epoch 10/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6460 - accuracy: 0.3738\n",
            "Epoch 11/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.6030 - accuracy: 0.3779\n",
            "Epoch 12/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.5734 - accuracy: 0.3859\n",
            "Epoch 13/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.5108 - accuracy: 0.4000\n",
            "Epoch 14/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.4870 - accuracy: 0.4082\n",
            "Epoch 15/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.4906 - accuracy: 0.4108\n",
            "Epoch 16/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.4355 - accuracy: 0.4426\n",
            "Epoch 17/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.4631 - accuracy: 0.4267\n",
            "Epoch 18/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.4403 - accuracy: 0.4351\n",
            "Epoch 19/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.4062 - accuracy: 0.4549\n",
            "Epoch 20/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3927 - accuracy: 0.4574\n",
            "Epoch 21/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3778 - accuracy: 0.4559\n",
            "Epoch 22/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3576 - accuracy: 0.4695\n",
            "Epoch 23/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3622 - accuracy: 0.4613\n",
            "Epoch 24/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3366 - accuracy: 0.4833\n",
            "Epoch 25/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3119 - accuracy: 0.4854\n",
            "Epoch 26/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3113 - accuracy: 0.4833\n",
            "Epoch 27/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2903 - accuracy: 0.4992\n",
            "Epoch 28/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.3087 - accuracy: 0.4895\n",
            "Epoch 29/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2768 - accuracy: 0.4964\n",
            "Epoch 30/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2823 - accuracy: 0.4956\n",
            "Epoch 31/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2419 - accuracy: 0.5105\n",
            "Epoch 32/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2360 - accuracy: 0.5105\n",
            "Epoch 33/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2164 - accuracy: 0.5172\n",
            "Epoch 34/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2447 - accuracy: 0.5069\n",
            "Epoch 35/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2453 - accuracy: 0.5062\n",
            "Epoch 36/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2139 - accuracy: 0.5213\n",
            "Epoch 37/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1974 - accuracy: 0.5277\n",
            "Epoch 38/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.2027 - accuracy: 0.5267\n",
            "Epoch 39/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1829 - accuracy: 0.5349\n",
            "Epoch 40/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1659 - accuracy: 0.5377\n",
            "Epoch 41/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1544 - accuracy: 0.5397\n",
            "Epoch 42/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1272 - accuracy: 0.5515\n",
            "Epoch 43/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1461 - accuracy: 0.5436\n",
            "Epoch 44/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1385 - accuracy: 0.5495\n",
            "Epoch 45/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1466 - accuracy: 0.5438\n",
            "Epoch 46/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0975 - accuracy: 0.5662\n",
            "Epoch 47/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.1178 - accuracy: 0.5533\n",
            "Epoch 48/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0938 - accuracy: 0.5574\n",
            "Epoch 49/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0664 - accuracy: 0.5695\n",
            "Epoch 50/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0669 - accuracy: 0.5687\n",
            "Epoch 51/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0393 - accuracy: 0.5777\n",
            "Epoch 52/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0442 - accuracy: 0.5749\n",
            "Epoch 53/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0456 - accuracy: 0.5795\n",
            "Epoch 54/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0417 - accuracy: 0.5851\n",
            "Epoch 55/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0573 - accuracy: 0.5772\n",
            "Epoch 56/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0358 - accuracy: 0.5808\n",
            "Epoch 57/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0177 - accuracy: 0.5910\n",
            "Epoch 58/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0368 - accuracy: 0.5792\n",
            "Epoch 59/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9883 - accuracy: 0.6003\n",
            "Epoch 60/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9807 - accuracy: 0.6005\n",
            "Epoch 61/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 1.0138 - accuracy: 0.5921\n",
            "Epoch 62/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9820 - accuracy: 0.5992\n",
            "Epoch 63/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9937 - accuracy: 0.6023\n",
            "Epoch 64/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9979 - accuracy: 0.6005\n",
            "Epoch 65/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9830 - accuracy: 0.6072\n",
            "Epoch 66/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9432 - accuracy: 0.6105\n",
            "Epoch 67/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9434 - accuracy: 0.6151\n",
            "Epoch 68/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9518 - accuracy: 0.6138\n",
            "Epoch 69/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9419 - accuracy: 0.6213\n",
            "Epoch 70/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9376 - accuracy: 0.6203\n",
            "Epoch 71/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9160 - accuracy: 0.6259\n",
            "Epoch 72/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9308 - accuracy: 0.6285\n",
            "Epoch 73/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9795 - accuracy: 0.6067\n",
            "Epoch 74/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9487 - accuracy: 0.6123\n",
            "Epoch 75/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9614 - accuracy: 0.6044\n",
            "Epoch 76/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9600 - accuracy: 0.6115\n",
            "Epoch 77/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9432 - accuracy: 0.6138\n",
            "Epoch 78/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9354 - accuracy: 0.6244\n",
            "Epoch 79/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8799 - accuracy: 0.6418\n",
            "Epoch 80/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9144 - accuracy: 0.6397\n",
            "Epoch 81/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9004 - accuracy: 0.6331\n",
            "Epoch 82/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8843 - accuracy: 0.6379\n",
            "Epoch 83/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8974 - accuracy: 0.6267\n",
            "Epoch 84/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9082 - accuracy: 0.6331\n",
            "Epoch 85/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9035 - accuracy: 0.6387\n",
            "Epoch 86/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9059 - accuracy: 0.6413\n",
            "Epoch 87/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8961 - accuracy: 0.6423\n",
            "Epoch 88/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9196 - accuracy: 0.6318\n",
            "Epoch 89/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8801 - accuracy: 0.6508\n",
            "Epoch 90/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8684 - accuracy: 0.6536\n",
            "Epoch 91/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8773 - accuracy: 0.6515\n",
            "Epoch 92/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8380 - accuracy: 0.6618\n",
            "Epoch 93/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8643 - accuracy: 0.6559\n",
            "Epoch 94/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8380 - accuracy: 0.6695\n",
            "Epoch 95/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8525 - accuracy: 0.6600\n",
            "Epoch 96/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8701 - accuracy: 0.6469\n",
            "Epoch 97/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8619 - accuracy: 0.6472\n",
            "Epoch 98/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8195 - accuracy: 0.6669\n",
            "Epoch 99/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8346 - accuracy: 0.6613\n",
            "Epoch 100/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8613 - accuracy: 0.6531\n",
            "Epoch 101/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8924 - accuracy: 0.6474\n",
            "Epoch 102/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8701 - accuracy: 0.6487\n",
            "Epoch 103/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.9040 - accuracy: 0.6423\n",
            "Epoch 104/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8578 - accuracy: 0.6462\n",
            "Epoch 105/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8284 - accuracy: 0.6638\n",
            "Epoch 106/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8492 - accuracy: 0.6538\n",
            "Epoch 107/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8596 - accuracy: 0.6513\n",
            "Epoch 108/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8466 - accuracy: 0.6626\n",
            "Epoch 109/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8758 - accuracy: 0.6515\n",
            "Epoch 110/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7824 - accuracy: 0.6831\n",
            "Epoch 111/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7693 - accuracy: 0.6879\n",
            "Epoch 112/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7807 - accuracy: 0.6874\n",
            "Epoch 113/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.8173 - accuracy: 0.6705\n",
            "Epoch 114/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7818 - accuracy: 0.6846\n",
            "Epoch 115/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7661 - accuracy: 0.6838\n",
            "Epoch 116/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7833 - accuracy: 0.6792\n",
            "Epoch 117/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7858 - accuracy: 0.6769\n",
            "Epoch 118/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7848 - accuracy: 0.6862\n",
            "Epoch 119/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7527 - accuracy: 0.6936\n",
            "Epoch 120/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7379 - accuracy: 0.7010\n",
            "Epoch 121/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7425 - accuracy: 0.7023\n",
            "Epoch 122/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7529 - accuracy: 0.6908\n",
            "Epoch 123/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7494 - accuracy: 0.6933\n",
            "Epoch 124/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7422 - accuracy: 0.6979\n",
            "Epoch 125/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7572 - accuracy: 0.6910\n",
            "Epoch 126/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7455 - accuracy: 0.7010\n",
            "Epoch 127/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7271 - accuracy: 0.7059\n",
            "Epoch 128/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7354 - accuracy: 0.6967\n",
            "Epoch 129/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7053 - accuracy: 0.7110\n",
            "Epoch 130/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7314 - accuracy: 0.7108\n",
            "Epoch 131/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7300 - accuracy: 0.7079\n",
            "Epoch 132/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6805 - accuracy: 0.7203\n",
            "Epoch 133/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7021 - accuracy: 0.7146\n",
            "Epoch 134/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7244 - accuracy: 0.7051\n",
            "Epoch 135/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7390 - accuracy: 0.7018\n",
            "Epoch 136/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7566 - accuracy: 0.6910\n",
            "Epoch 137/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7283 - accuracy: 0.7021\n",
            "Epoch 138/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7485 - accuracy: 0.7031\n",
            "Epoch 139/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7720 - accuracy: 0.6962\n",
            "Epoch 140/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7589 - accuracy: 0.6982\n",
            "Epoch 141/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7118 - accuracy: 0.7095\n",
            "Epoch 142/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7712 - accuracy: 0.6910\n",
            "Epoch 143/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6669 - accuracy: 0.7256\n",
            "Epoch 144/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6239 - accuracy: 0.7469\n",
            "Epoch 145/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7388 - accuracy: 0.6982\n",
            "Epoch 146/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6709 - accuracy: 0.7244\n",
            "Epoch 147/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6674 - accuracy: 0.7228\n",
            "Epoch 148/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6765 - accuracy: 0.7221\n",
            "Epoch 149/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6953 - accuracy: 0.7192\n",
            "Epoch 150/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6795 - accuracy: 0.7208\n",
            "Epoch 151/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6673 - accuracy: 0.7297\n",
            "Epoch 152/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6607 - accuracy: 0.7264\n",
            "Epoch 153/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6974 - accuracy: 0.7123\n",
            "Epoch 154/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6854 - accuracy: 0.7213\n",
            "Epoch 155/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6344 - accuracy: 0.7397\n",
            "Epoch 156/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6736 - accuracy: 0.7323\n",
            "Epoch 157/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6158 - accuracy: 0.7436\n",
            "Epoch 158/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7288 - accuracy: 0.7136\n",
            "Epoch 159/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6595 - accuracy: 0.7285\n",
            "Epoch 160/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7800 - accuracy: 0.6803\n",
            "Epoch 161/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6880 - accuracy: 0.7238\n",
            "Epoch 162/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6518 - accuracy: 0.7351\n",
            "Epoch 163/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6660 - accuracy: 0.7292\n",
            "Epoch 164/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6316 - accuracy: 0.7349\n",
            "Epoch 165/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6169 - accuracy: 0.7449\n",
            "Epoch 166/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5927 - accuracy: 0.7523\n",
            "Epoch 167/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6350 - accuracy: 0.7364\n",
            "Epoch 168/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7078 - accuracy: 0.7126\n",
            "Epoch 169/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6417 - accuracy: 0.7464\n",
            "Epoch 170/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.7203 - accuracy: 0.7085\n",
            "Epoch 171/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6290 - accuracy: 0.7387\n",
            "Epoch 172/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6099 - accuracy: 0.7446\n",
            "Epoch 173/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6444 - accuracy: 0.7415\n",
            "Epoch 174/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5988 - accuracy: 0.7485\n",
            "Epoch 175/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5698 - accuracy: 0.7664\n",
            "Epoch 176/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5877 - accuracy: 0.7600\n",
            "Epoch 177/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5745 - accuracy: 0.7636\n",
            "Epoch 178/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6127 - accuracy: 0.7515\n",
            "Epoch 179/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6139 - accuracy: 0.7485\n",
            "Epoch 180/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5368 - accuracy: 0.7746\n",
            "Epoch 181/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5985 - accuracy: 0.7554\n",
            "Epoch 182/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5948 - accuracy: 0.7508\n",
            "Epoch 183/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5864 - accuracy: 0.7579\n",
            "Epoch 184/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6138 - accuracy: 0.7549\n",
            "Epoch 185/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6393 - accuracy: 0.7377\n",
            "Epoch 186/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5991 - accuracy: 0.7549\n",
            "Epoch 187/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5606 - accuracy: 0.7667\n",
            "Epoch 188/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5313 - accuracy: 0.7769\n",
            "Epoch 189/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5720 - accuracy: 0.7664\n",
            "Epoch 190/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5656 - accuracy: 0.7746\n",
            "Epoch 191/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5701 - accuracy: 0.7577\n",
            "Epoch 192/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5405 - accuracy: 0.7754\n",
            "Epoch 193/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5267 - accuracy: 0.7890\n",
            "Epoch 194/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.6533 - accuracy: 0.7346\n",
            "Epoch 195/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5449 - accuracy: 0.7767\n",
            "Epoch 196/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5604 - accuracy: 0.7723\n",
            "Epoch 197/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5421 - accuracy: 0.7751\n",
            "Epoch 198/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5186 - accuracy: 0.7833\n",
            "Epoch 199/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5493 - accuracy: 0.7751\n",
            "Epoch 200/200\n",
            "122/122 [==============================] - 0s 3ms/step - loss: 0.5930 - accuracy: 0.7567\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "cellView": "both",
        "id": "x5DX2gq7oA-O",
        "colab": {}
      },
      "source": [
        "#Guardar\n",
        "#@markdown optimizer_epochs_samples_features_accuracy_version\n",
        "\n",
        "nombre = \"DNN_SGD_200e_650m_2f_0.75_V1\" #@param {type:\"string\"}\n",
        "model.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.h5')\n",
        "np.save('/content/drive/My Drive/IA2/DATA/Models/model_'+nombre+'.npy', le.classes_)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "LVfnoti7URp8",
        "colab": {}
      },
      "source": [
        "# Cargar modelo\n",
        "model = tf.keras.models.load_model('/content/drive/My Drive/IA2/DATA/Models/model_DNN_SGD_200e_650m_2f_0.75_V1.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "9Irs-lyUTZ1L",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "e631db49-9e2d-4d8e-8dcf-ec6a66d742fd"
      },
      "source": [
        "#@title Predicción\n",
        "X_test = x_test.reshape(x_test.shape[0],x_test.shape[1],1)\n",
        "y_predict = np.argmax(model.predict(X_test), axis=1)\n",
        "y_test_single = np.argmax(y_test, axis=1)\n",
        "conf_matrix = confusion_matrix(y_test_single, y_predict)\n",
        "print(conf_matrix)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[ 16  13   8  11  92   1]\n",
            " [  4  83   8  39  49   5]\n",
            " [  4  24  31   4 103   2]\n",
            " [  1  32   5  73  40  11]\n",
            " [ 13  53  22  48 335  17]\n",
            " [  4  15   0  24  36  74]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "4j8TyAtbTZ1T",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        },
        "outputId": "379189f7-e6fb-4313-8c80-3252a7132bc6"
      },
      "source": [
        "text = [\"ON\", \"OFF\", \"YES\", \"NO\", \"STOP\", \"UNKNOWN\"]\n",
        "#Vector para gráficas\n",
        "DNN_SGD=[]\n",
        "for i in range(len(text)):\n",
        "  acierto_promedio = conf_matrix[i][i]/(np.sum(conf_matrix[i], axis=0))\n",
        "  DNN_SGD.append(np.round(acierto_promedio, 2))\n",
        "  print(f\"Aciertos clase {text[i]}: {np.round(acierto_promedio*100, 2)}%\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Aciertos clase ON: 11.35%\n",
            "Aciertos clase OFF: 44.15%\n",
            "Aciertos clase YES: 18.45%\n",
            "Aciertos clase NO: 45.06%\n",
            "Aciertos clase STOP: 68.65%\n",
            "Aciertos clase UNKNOWN: 48.37%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "GoorHbWsT3nA"
      },
      "source": [
        "### Resultados con DNN"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "cellView": "both",
        "id": "WqbyQFb2T3nB",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "673fde53-e5c2-4d77-d245-2799fa455c1b"
      },
      "source": [
        "barWidth=0.2\n",
        "plt.bar(np.arange(len(DNN_ADAM)), DNN_ADAM, color=\"#6A5ACD\", width=barWidth, label=\"Adam\");\n",
        "plt.bar([x + barWidth for x in np.arange(len(DNN_ADAM))], DNN_SGD, color=\"#00BFFF\", width=barWidth, label=\"SGD\");\n",
        "plt.xlabel(\"Clases\");\n",
        "plt.ylabel(\"Aciertos en %\");\n",
        "plt.xticks([r + barWidth for r in range(len(text))], text);\n",
        "plt.title(\"Resultados de la implementación con modelo DNN\");\n",
        "plt.legend();\n",
        "plt.show();"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LbmqubBxg4QR",
        "colab_type": "text"
      },
      "source": [
        "#4. IMPLEMENTACIÓN\n",
        "Para realizar pruebas reales con los modelos entrenados se desarrolla una interfaz compatible con el entorno del notebook. Esta implementación permite grabar una frase con el micrófono y posteriormente muestra un conteo de los comandos predichos."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "R9GWE46tDXDD",
        "colab_type": "code",
        "cellView": "both",
        "colab": {}
      },
      "source": [
        "#@title Cargar modelo preentrenado\n",
        "modelo = 'DNN_Adamax_200e_650m_2f_0.87_V1' #@param ['RNN_SGD_50e_650m_2f_0.76_V1', 'LSTM_SGD_50e_650m_2f_0.86_V1', 'GRU_SGD_50e_650m_2f_1_V1', 'DNN_Adamax_200e_650m_2f_0.87_V1']\n",
        "modelo_prueba = tf.keras.models.load_model('Models/model_'+modelo+'.h5')\n",
        "encoder = LabelEncoder()\n",
        "encoder.classes_ = np.load('Models/model_'+modelo+'.npy')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "cellView": "form",
        "id": "jP4bLKsgYk2N",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 511
        },
        "outputId": "3a5ae25c-97cb-4063-f851-30991c4a2e42"
      },
      "source": [
        "#@title .\n",
        "display(IPython.display.HTML('''\n",
        "<html>\n",
        "  <head>\n",
        "    <link rel=\"stylesheet\" href=\"https://stackpath.bootstrapcdn.com/bootstrap/4.5.0/css/bootstrap.min.css\"\n",
        "        integrity=\"sha384-9aIt2nRpC12Uk9gS9baDl411NQApFmC26EwAOH8WgZl5MYYxFfc+NcPb1dKGj7Sk\" crossorigin=\"anonymous\">\n",
        "    </haed>\n",
        "    <body>\n",
        "      <div class=\"container\">  \n",
        "        <div class=\"row\">\n",
        "          <img src=\"https://drive.google.com/uc?id=1mVJfkz8r5vcVDGcmzos2f6BF5m9MiK8N\" class=\"card-img-top\" alt=\"banner\">\n",
        "        </div>   \n",
        "        <div class=\"row\">\n",
        "          <div class=\"col-3 bg-secondary text-center text-white\">\n",
        "            <h4 class=\"text-white mt-4\">RECORD COMMAND</h4>\n",
        "            <audio id = \"audio\" style=\"width: 100%;\" controls></audio>\n",
        "            <div id=\"screen\" style=\"font-family: Calibri,Arial;font-weight: 300;font-size: 45px;letter-spacing: 3px;\">\n",
        "              00:00\n",
        "            </div>\n",
        "            <button id='button' type=\"button\" class=\"btn btn-primary mx-auto d-block mb-3\">Record</button>\n",
        "          </div>\n",
        "          <div class=\"col-9\" style=\"background-color: #E0E0E0; padding: 0px !important\">\n",
        "            <div class=\"container\">\n",
        "              <div class=\"row\">\n",
        "                <div class=\"p-1 w-100 text-center\" style=\"background-color: #14B9CF;font-weight: 600;font-size: 20px;\" >\n",
        "                  Command: <span id=\"resultado\">\n",
        "                  <div class=\"spinner-border\" role=\"status\">\n",
        "                    <span class=\"sr-only\">Loading...</span>\n",
        "                  </div>\n",
        "                  <span>\n",
        "                </div>\n",
        "              </div>\n",
        "              <div class=\"row\">\n",
        "                <div class=\"col-sm\">\n",
        "                  <img  width=\"100%\" id=\"imagen1\" src=\"https://drive.google.com/uc?id=1d92jKlIJL-ezrLOAtpsK2dIE6cxK7MVO\" />\n",
        "                </div>\n",
        "                <div class=\"col-sm\">\n",
        "                  <img  width=\"100%\" id=\"imagen2\" src=\"https://drive.google.com/uc?id=19BeoY0biKIDCMej59ECN3unxrc8uwk09\" />\n",
        "                </div>\n",
        "                <div class=\"col-sm\">\n",
        "                  <img  width=\"100%\" id=\"imagen3\" src=\"https://drive.google.com/uc?id=11-M5rIg0ifCtnL1K4tGvyRfwyaeO7VSF\" />\n",
        "                </div>\n",
        "              </div>\n",
        "              <div class=\"row text-center p-2\" style=\"background-color: #c4c4c4;font-weight: 600;font-size: 16px;\">\n",
        "                <div class=\"col\">\n",
        "                    On <span id=\"on\" class=\"badge badge-light\">0</span>\n",
        "                </div>\n",
        "                <div class=\"col\">\n",
        "                    Off <span id=\"off\" class=\"badge badge-light\">0</span>\n",
        "                </div>\n",
        "                <div class=\"col\">\n",
        "                    Stop <span id=\"stop\" class=\"badge badge-light\">0</span>\n",
        "                </div>\n",
        "                <div class=\"col\">\n",
        "                    Yes <span id=\"yes\" class=\"badge badge-light\">0</span>\n",
        "                </div>\n",
        "                <div class=\"col\">\n",
        "                    No <span id=\"no\" class=\"badge badge-light\">0</span>\n",
        "                </div>\n",
        "                <div class=\"col\">\n",
        "                    Unknown <span id=\"unknown\" class=\"badge badge-light\">0</span>\n",
        "                </div>\n",
        "              </div>\n",
        "            </div>\n",
        "          </div>\n",
        "        </div>\n",
        "      </div>\n",
        "    </body>\n",
        "\n",
        "    <script>\n",
        "      var isMarch = false; \n",
        "      var acumularTime = 0; \n",
        "      var screen = document.getElementById(\"screen\")\n",
        "      var resultado = document.getElementById(\"resultado\")\n",
        "      // var grab = document.getElementById(\"grabacion\")\n",
        "      var imagen1 = document.getElementById(\"imagen1\")\n",
        "      var imagen2 = document.getElementById(\"imagen2\")\n",
        "      var imagen3 = document.getElementById(\"imagen3\")\n",
        "      var b_on = document.getElementById(\"on\")\n",
        "      var b_off = document.getElementById(\"off\")\n",
        "      var b_stop = document.getElementById(\"stop\")\n",
        "      var b_yes = document.getElementById(\"yes\")\n",
        "      var b_no = document.getElementById(\"no\")\n",
        "      var b_unknown = document.getElementById(\"unknown\")\n",
        "      var button = document.getElementById(\"button\")\n",
        "      // var tabla = document.getElementById(\"tabla\")\n",
        "      // var load = document.getElementById(\"load\")\n",
        "      var start_rec = false;\n",
        "      // tabla.style.display = \"none\";\n",
        "\n",
        "      function cronometro () { \n",
        "        timeActual = new Date();\n",
        "        acumularTime = timeActual - timeInicial;\n",
        "        acumularTime2 = new Date();\n",
        "        acumularTime2.setTime(acumularTime); \n",
        "        cc = Math.round(acumularTime2.getMilliseconds()/10);\n",
        "        ss = acumularTime2.getSeconds();\n",
        "        if (cc < 10) {cc = \"0\"+cc;}\n",
        "        if (ss < 10) {ss = \"0\"+ss;} \n",
        "        screen.innerHTML = ss+\":\"+cc;\n",
        "      }\n",
        "\n",
        "      function start() {\n",
        "        if (isMarch == false) { \n",
        "          timeInicial = new Date();\n",
        "          control = setInterval(cronometro,10);\n",
        "          isMarch = true;\n",
        "        }\n",
        "      }\n",
        "\n",
        "      function stop() { \n",
        "        if (isMarch == true) {\n",
        "          clearInterval(control);\n",
        "          isMarch = false;\n",
        "        }     \n",
        "      }   \n",
        "      \n",
        "      function toggleRecording() {\n",
        "        if (recorder && recorder.state == \"recording\") {\n",
        "            recorder.stop();\n",
        "            stop();\n",
        "            gumStream.getAudioTracks()[0].stop();\n",
        "        }\n",
        "      }   \n",
        "      async function send(base64data) {\n",
        "        var c_on = 0\n",
        "        var c_off = 0\n",
        "        var c_yes = 0\n",
        "        var c_no = 0\n",
        "        var c_stop = 0\n",
        "        var c_unknown = 0\n",
        "        const result = await google.colab.kernel.invokeFunction('notebook.sendAudio', [base64data.toString()] , {});\n",
        "        const predicts = result.data['application/json'].result\n",
        "        resultado.innerHTML = predicts.toString()\n",
        "        predicts.forEach(function (predict){\n",
        "          console.log(predict)\n",
        "          if(predict == 'on'){\n",
        "            imagen1.src = \"https://drive.google.com/uc?id=1H3YbVnvHhL7KiR9f_QNPgZu02Wm2b_Fd\";\n",
        "            c_on+=1\n",
        "          }\n",
        "          if(predict == 'off'){\n",
        "            imagen1.src = \"https://drive.google.com/uc?id=1d92jKlIJL-ezrLOAtpsK2dIE6cxK7MVO\";\n",
        "            c_off+=1\n",
        "          }\n",
        "          if(predict == 'no'){\n",
        "            imagen3.src = \"https://drive.google.com/uc?id=11-M5rIg0ifCtnL1K4tGvyRfwyaeO7VSF\";\n",
        "            c_no+=1\n",
        "          }\n",
        "          if(predict == 'yes'){\n",
        "            imagen3.src = \"https://drive.google.com/uc?id=1R7szogaVmuGFXViuqFYocfLYhO61T5X9\";\n",
        "            c_yes+=1\n",
        "          }\n",
        "          if(predict == 'stop'){\n",
        "            imagen2.src = \"https://drive.google.com/uc?id=19X3uT8VUZAYQXgMGs8R-ty1ifZj9IuHM\";\n",
        "            c_stop+=1\n",
        "          }\n",
        "          if(predict == 'unknow'){\n",
        "            c_unknown+=1\n",
        "          }\n",
        "          b_no.innerHTML = c_no\n",
        "          b_yes.innerHTML = c_yes\n",
        "          b_on.innerHTML = c_on\n",
        "          b_off.innerHTML = c_off\n",
        "          b_stop.innerHTML = c_stop\n",
        "          b_unknown.innerHTML = c_unknown\n",
        "        })\n",
        "        // \n",
        "      }\n",
        "\n",
        "      var handleSuccess = function(stream) {\n",
        "        gumStream = stream;\n",
        "        var options = {\n",
        "          //bitsPerSecond: 8000, //chrome seems to ignore, always 48k\n",
        "          mimeType : 'audio/webm;codecs=opus'\n",
        "          //mimeType : 'audio/webm;codecs=pcm'\n",
        "        };            \n",
        "        recorder = new MediaRecorder(stream);\n",
        "        recorder.ondataavailable = function(e) {            \n",
        "          var url = URL.createObjectURL(e.data);\n",
        "          var preview = document.getElementById('audio');\n",
        "          preview.src = url;\n",
        "\n",
        "          reader = new FileReader();\n",
        "          reader.readAsDataURL(e.data); \n",
        "          reader.onloadend = function() {\n",
        "            base64data = reader.result;\n",
        "            send(base64data);\n",
        "          }\n",
        "        };\n",
        "        recorder.start();\n",
        "      };\n",
        "\n",
        "      function grabar() {\n",
        "        button.innerText = \"Stop\"\n",
        "        imagen2.src = \"https://drive.google.com/uc?id=19BeoY0biKIDCMej59ECN3unxrc8uwk09\";\n",
        "        button.classList.remove(\"btn-primary\");\n",
        "        button.classList.add(\"btn-danger\");\n",
        "\n",
        "        navigator.mediaDevices.getUserMedia({audio: true}).then(handleSuccess);\n",
        "        start();\n",
        "      }\n",
        "\n",
        "      document.querySelector('#button').onclick = () => {\n",
        "        if (start_rec){\n",
        "          button.innerText = \"Record\"\n",
        "          button.classList.remove(\"btn-danger\");\n",
        "          button.classList.add(\"btn-primary\");\n",
        "          toggleRecording()\n",
        "          start_rec = false\n",
        "        }else{\n",
        "          // load.style.display = \"none\";\n",
        "          // tabla.style.display = \"block\";\n",
        "          resultado.innerHTML = '<div class=\"spinner-border\" role=\"status\"><span class=\"sr-only\">Loading...</span></div>'\n",
        "          // document.querySelector('#youtube-video').contentWindow.postMessage('{\"event\":\"command\",\"func\":\"' + 'playVideo' + '\",\"args\":\"\"}', '*');\n",
        "          grabar()\n",
        "          b_no.innerHTML = 0\n",
        "          b_yes.innerHTML = 0\n",
        "          b_on.innerHTML = 0\n",
        "          b_off.innerHTML = 0\n",
        "          b_stop.innerHTML = 0\n",
        "          b_unknown.innerHTML = 0\n",
        "          start_rec = true\n",
        "        }\n",
        "      };\n",
        "    </script>\n",
        "    </html>\n",
        "    '''))\n",
        "\n",
        "def sendAudio(base64audio):\n",
        "\n",
        "  binary = b64decode(base64audio.split(',')[1])\n",
        "  \n",
        "  process = (ffmpeg\n",
        "    .input('pipe:0')\n",
        "    .output('pipe:1', format='wav')\n",
        "    .run_async(pipe_stdin=True, pipe_stdout=True, pipe_stderr=True, quiet=True, overwrite_output=True)\n",
        "  )\n",
        "  output, err = process.communicate(input=binary)\n",
        "  \n",
        "  riff_chunk_size = len(output) - 8\n",
        "  # Break up the chunk size into four bytes, held in b.\n",
        "  q = riff_chunk_size\n",
        "  b = []\n",
        "  for i in range(4):\n",
        "      q, r = divmod(q, 256)\n",
        "      b.append(r)\n",
        "\n",
        "  # Replace bytes 4:8 in proc.stdout with the actual size of the RIFF chunk.\n",
        "  riff = output[:4] + bytes(b) + output[8:]\n",
        "\n",
        "  sr, audio = wav_read(io.BytesIO(riff))\n",
        "  now = datetime.now()\n",
        "  name = now.strftime('%d_%m_%Y_%H_%M_%S')\n",
        "  scipy.io.wavfile.write('/content/drive/My Drive/IA2/DATA/Recs/rec_'+name+'.wav', sr, audio)\n",
        "  \n",
        "  name_file = '/content/drive/My Drive/IA2/DATA/Recs/rec_'+str(name)+'.wav'\n",
        "  sound_file = AudioSegment.from_wav(name_file)\n",
        "  audio_chunks = split_on_silence(sound_file, min_silence_len=100,silence_thresh=-25)\n",
        "  result = []\n",
        "\n",
        "  for i, chunk in enumerate(audio_chunks):\n",
        "    out_file = \"/content/drive/My Drive/IA2/DATA/Recs/chunk{0}.wav\".format(i)\n",
        "    chunk.export(out_file, format=\"wav\")\n",
        "\n",
        "  for i, chunk in enumerate(audio_chunks):\n",
        "    # out_file = \"/content/drive/My Drive/IA2/DATA/Recs/chunk{0}.wav\".format(i)\n",
        "    # chunk.export(out_file, format=\"wav\")\n",
        "    file_name_rec = \"chunk{0}.wav\".format(i)\n",
        "    rec_audio = extract_features('Recs/', file_name_rec)\n",
        "    chroma_audio  = extract_features_chroma('Recs/', file_name_rec)\n",
        "    x_audio = np.concatenate((rec_audio, chroma_audio)).reshape(1,92,1)\n",
        "    # Si es RNN, aplicar model.predict\n",
        "    if \"RNN\" or \"DNN\" in modelo:\n",
        "      predict_rec = modelo_prueba.predict(x_audio)\n",
        "      p = encoder.inverse_transform(np.argmax(predict_rec, axis=1))[0]\n",
        "    else:\n",
        "      predict_rec = modelo_prueba.predict_classes(x_audio)\n",
        "      p = encoder.inverse_transform(predict_rec)[0]\n",
        "    \n",
        "    result.append(p)\n",
        "\n",
        "  return IPython.display.JSON({'result': result})\n",
        "\n",
        "\n",
        "output.register_callback('notebook.sendAudio', sendAudio)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "<html>\n",
              "  <head>\n",
              "    <link rel=\"stylesheet\" href=\"https://stackpath.bootstrapcdn.com/bootstrap/4.5.0/css/bootstrap.min.css\"\n",
              "        integrity=\"sha384-9aIt2nRpC12Uk9gS9baDl411NQApFmC26EwAOH8WgZl5MYYxFfc+NcPb1dKGj7Sk\" crossorigin=\"anonymous\">\n",
              "    </haed>\n",
              "    <body>\n",
              "      <div class=\"container\">  \n",
              "        <div class=\"row\">\n",
              "          <img src=\"https://drive.google.com/uc?id=1mVJfkz8r5vcVDGcmzos2f6BF5m9MiK8N\" class=\"card-img-top\" alt=\"banner\">\n",
              "        </div>   \n",
              "        <div class=\"row\">\n",
              "          <div class=\"col-3 bg-secondary text-center text-white\">\n",
              "            <h4 class=\"text-white mt-4\">RECORD COMMAND</h4>\n",
              "            <audio id = \"audio\" style=\"width: 100%;\" controls></audio>\n",
              "            <div id=\"screen\" style=\"font-family: Calibri,Arial;font-weight: 300;font-size: 45px;letter-spacing: 3px;\">\n",
              "              00:00\n",
              "            </div>\n",
              "            <button id='button' type=\"button\" class=\"btn btn-primary mx-auto d-block mb-3\">Record</button>\n",
              "          </div>\n",
              "          <div class=\"col-9\" style=\"background-color: #E0E0E0; padding: 0px !important\">\n",
              "            <div class=\"container\">\n",
              "              <div class=\"row\">\n",
              "                <div class=\"p-1 w-100 text-center\" style=\"background-color: #14B9CF;font-weight: 600;font-size: 20px;\" >\n",
              "                  Command: <span id=\"resultado\">\n",
              "                  <div class=\"spinner-border\" role=\"status\">\n",
              "                    <span class=\"sr-only\">Loading...</span>\n",
              "                  </div>\n",
              "                  <span>\n",
              "                </div>\n",
              "              </div>\n",
              "              <div class=\"row\">\n",
              "                <div class=\"col-sm\">\n",
              "                  <img  width=\"100%\" id=\"imagen1\" src=\"https://drive.google.com/uc?id=1d92jKlIJL-ezrLOAtpsK2dIE6cxK7MVO\" />\n",
              "                </div>\n",
              "                <div class=\"col-sm\">\n",
              "                  <img  width=\"100%\" id=\"imagen2\" src=\"https://drive.google.com/uc?id=19BeoY0biKIDCMej59ECN3unxrc8uwk09\" />\n",
              "                </div>\n",
              "                <div class=\"col-sm\">\n",
              "                  <img  width=\"100%\" id=\"imagen3\" src=\"https://drive.google.com/uc?id=11-M5rIg0ifCtnL1K4tGvyRfwyaeO7VSF\" />\n",
              "                </div>\n",
              "              </div>\n",
              "              <div class=\"row text-center p-2\" style=\"background-color: #c4c4c4;font-weight: 600;font-size: 16px;\">\n",
              "                <div class=\"col\">\n",
              "                    On <span id=\"on\" class=\"badge badge-light\">0</span>\n",
              "                </div>\n",
              "                <div class=\"col\">\n",
              "                    Off <span id=\"off\" class=\"badge badge-light\">0</span>\n",
              "                </div>\n",
              "                <div class=\"col\">\n",
              "                    Stop <span id=\"stop\" class=\"badge badge-light\">0</span>\n",
              "                </div>\n",
              "                <div class=\"col\">\n",
              "                    Yes <span id=\"yes\" class=\"badge badge-light\">0</span>\n",
              "                </div>\n",
              "                <div class=\"col\">\n",
              "                    No <span id=\"no\" class=\"badge badge-light\">0</span>\n",
              "                </div>\n",
              "                <div class=\"col\">\n",
              "                    Unknown <span id=\"unknown\" class=\"badge badge-light\">0</span>\n",
              "                </div>\n",
              "              </div>\n",
              "            </div>\n",
              "          </div>\n",
              "        </div>\n",
              "      </div>\n",
              "    </body>\n",
              "\n",
              "    <script>\n",
              "      var isMarch = false; \n",
              "      var acumularTime = 0; \n",
              "      var screen = document.getElementById(\"screen\")\n",
              "      var resultado = document.getElementById(\"resultado\")\n",
              "      // var grab = document.getElementById(\"grabacion\")\n",
              "      var imagen1 = document.getElementById(\"imagen1\")\n",
              "      var imagen2 = document.getElementById(\"imagen2\")\n",
              "      var imagen3 = document.getElementById(\"imagen3\")\n",
              "      var b_on = document.getElementById(\"on\")\n",
              "      var b_off = document.getElementById(\"off\")\n",
              "      var b_stop = document.getElementById(\"stop\")\n",
              "      var b_yes = document.getElementById(\"yes\")\n",
              "      var b_no = document.getElementById(\"no\")\n",
              "      var b_unknown = document.getElementById(\"unknown\")\n",
              "      var button = document.getElementById(\"button\")\n",
              "      // var tabla = document.getElementById(\"tabla\")\n",
              "      // var load = document.getElementById(\"load\")\n",
              "      var start_rec = false;\n",
              "      // tabla.style.display = \"none\";\n",
              "\n",
              "      function cronometro () { \n",
              "        timeActual = new Date();\n",
              "        acumularTime = timeActual - timeInicial;\n",
              "        acumularTime2 = new Date();\n",
              "        acumularTime2.setTime(acumularTime); \n",
              "        cc = Math.round(acumularTime2.getMilliseconds()/10);\n",
              "        ss = acumularTime2.getSeconds();\n",
              "        if (cc < 10) {cc = \"0\"+cc;}\n",
              "        if (ss < 10) {ss = \"0\"+ss;} \n",
              "        screen.innerHTML = ss+\":\"+cc;\n",
              "      }\n",
              "\n",
              "      function start() {\n",
              "        if (isMarch == false) { \n",
              "          timeInicial = new Date();\n",
              "          control = setInterval(cronometro,10);\n",
              "          isMarch = true;\n",
              "        }\n",
              "      }\n",
              "\n",
              "      function stop() { \n",
              "        if (isMarch == true) {\n",
              "          clearInterval(control);\n",
              "          isMarch = false;\n",
              "        }     \n",
              "      }   \n",
              "      \n",
              "      function toggleRecording() {\n",
              "        if (recorder && recorder.state == \"recording\") {\n",
              "            recorder.stop();\n",
              "            stop();\n",
              "            gumStream.getAudioTracks()[0].stop();\n",
              "        }\n",
              "      }   \n",
              "      async function send(base64data) {\n",
              "        var c_on = 0\n",
              "        var c_off = 0\n",
              "        var c_yes = 0\n",
              "        var c_no = 0\n",
              "        var c_stop = 0\n",
              "        var c_unknown = 0\n",
              "        const result = await google.colab.kernel.invokeFunction('notebook.sendAudio', [base64data.toString()] , {});\n",
              "        const predicts = result.data['application/json'].result\n",
              "        resultado.innerHTML = predicts.toString()\n",
              "        predicts.forEach(function (predict){\n",
              "          console.log(predict)\n",
              "          if(predict == 'on'){\n",
              "            imagen1.src = \"https://drive.google.com/uc?id=1H3YbVnvHhL7KiR9f_QNPgZu02Wm2b_Fd\";\n",
              "            c_on+=1\n",
              "          }\n",
              "          if(predict == 'off'){\n",
              "            imagen1.src = \"https://drive.google.com/uc?id=1d92jKlIJL-ezrLOAtpsK2dIE6cxK7MVO\";\n",
              "            c_off+=1\n",
              "          }\n",
              "          if(predict == 'no'){\n",
              "            imagen3.src = \"https://drive.google.com/uc?id=11-M5rIg0ifCtnL1K4tGvyRfwyaeO7VSF\";\n",
              "            c_no+=1\n",
              "          }\n",
              "          if(predict == 'yes'){\n",
              "            imagen3.src = \"https://drive.google.com/uc?id=1R7szogaVmuGFXViuqFYocfLYhO61T5X9\";\n",
              "            c_yes+=1\n",
              "          }\n",
              "          if(predict == 'stop'){\n",
              "            imagen2.src = \"https://drive.google.com/uc?id=19X3uT8VUZAYQXgMGs8R-ty1ifZj9IuHM\";\n",
              "            c_stop+=1\n",
              "          }\n",
              "          if(predict == 'unknow'){\n",
              "            c_unknown+=1\n",
              "          }\n",
              "          b_no.innerHTML = c_no\n",
              "          b_yes.innerHTML = c_yes\n",
              "          b_on.innerHTML = c_on\n",
              "          b_off.innerHTML = c_off\n",
              "          b_stop.innerHTML = c_stop\n",
              "          b_unknown.innerHTML = c_unknown\n",
              "        })\n",
              "        // \n",
              "      }\n",
              "\n",
              "      var handleSuccess = function(stream) {\n",
              "        gumStream = stream;\n",
              "        var options = {\n",
              "          //bitsPerSecond: 8000, //chrome seems to ignore, always 48k\n",
              "          mimeType : 'audio/webm;codecs=opus'\n",
              "          //mimeType : 'audio/webm;codecs=pcm'\n",
              "        };            \n",
              "        recorder = new MediaRecorder(stream);\n",
              "        recorder.ondataavailable = function(e) {            \n",
              "          var url = URL.createObjectURL(e.data);\n",
              "          var preview = document.getElementById('audio');\n",
              "          preview.src = url;\n",
              "\n",
              "          reader = new FileReader();\n",
              "          reader.readAsDataURL(e.data); \n",
              "          reader.onloadend = function() {\n",
              "            base64data = reader.result;\n",
              "            send(base64data);\n",
              "          }\n",
              "        };\n",
              "        recorder.start();\n",
              "      };\n",
              "\n",
              "      function grabar() {\n",
              "        button.innerText = \"Stop\"\n",
              "        imagen2.src = \"https://drive.google.com/uc?id=19BeoY0biKIDCMej59ECN3unxrc8uwk09\";\n",
              "        button.classList.remove(\"btn-primary\");\n",
              "        button.classList.add(\"btn-danger\");\n",
              "\n",
              "        navigator.mediaDevices.getUserMedia({audio: true}).then(handleSuccess);\n",
              "        start();\n",
              "      }\n",
              "\n",
              "      document.querySelector('#button').onclick = () => {\n",
              "        if (start_rec){\n",
              "          button.innerText = \"Record\"\n",
              "          button.classList.remove(\"btn-danger\");\n",
              "          button.classList.add(\"btn-primary\");\n",
              "          toggleRecording()\n",
              "          start_rec = false\n",
              "        }else{\n",
              "          // load.style.display = \"none\";\n",
              "          // tabla.style.display = \"block\";\n",
              "          resultado.innerHTML = '<div class=\"spinner-border\" role=\"status\"><span class=\"sr-only\">Loading...</span></div>'\n",
              "          // document.querySelector('#youtube-video').contentWindow.postMessage('{\"event\":\"command\",\"func\":\"' + 'playVideo' + '\",\"args\":\"\"}', '*');\n",
              "          grabar()\n",
              "          b_no.innerHTML = 0\n",
              "          b_yes.innerHTML = 0\n",
              "          b_on.innerHTML = 0\n",
              "          b_off.innerHTML = 0\n",
              "          b_stop.innerHTML = 0\n",
              "          b_unknown.innerHTML = 0\n",
              "          start_rec = true\n",
              "        }\n",
              "      };\n",
              "    </script>\n",
              "    </html>\n",
              "    "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gwIbK4SqxFEU",
        "colab_type": "text"
      },
      "source": [
        "# 5. CONCLUSIONES"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "y_NqPWgnVaaE",
        "colab_type": "text"
      },
      "source": [
        "* Las características frecuenciales MFCC y Chroma, en conjunto, son una representación que garantiza una predicción aceptable de los comandos de voz. Se evidenció que el ruido influye significativamente en los resultados debido a la alta sensibilidad de los micrófonos convencionales.\n",
        "\n",
        "* La detección de comandos en una frase es un reto de mayor complejidad en comparación a la detención de comandos por palabras, ya que el tratamiento de la frase captada depende de su duración y nivel de potencia para poder identificar cada palabra en esta.\n",
        "\n",
        "* Una arquitectura sencilla como la DNN puede lograr un rendimiento igual o mayor de óptimo con respecto a las redes neuronales recurrentes. En las pruebas reales, el modelo con más aciertos es el DNN (Adamax). Por otro lado, para las pruebas con la partición test del dataset, el modelo GRU y RNN simple poseen una mayor precisión"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oEmCPG050OeL",
        "colab_type": "text"
      },
      "source": [
        "# Referencias Bibliográficas:"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1u7Al0wAVbCW",
        "colab_type": "text"
      },
      "source": [
        "* Bonafonte, A., (2016). El Deep Learning Revoluciona Las Tecnologías Del Habla | Blog CIT UPC. (2016). Consultado el 6 de Septiembre de 2020, en https://blog.cit.upc.edu/?p=965. \n",
        "\n",
        "* TensorFlow Speech Recognition Challenge | Kaggle. (2016). Consultado el 6 de Agosto de 2020, en https://www.kaggle.com/c/tensorflow-speech-recognition-challenge \n",
        "\n",
        "* Criptografía práctica. (2013). Consultado el 10 de agosto de 2020, en http://practicalcryptography.com/miscellaneous/machine-learning/guide-mel-frequency-cepstral-coefficients-mfccs/\n",
        " \n",
        "* Jogy, J. (2019). How I Understood: What features to consider while training audio files?. Consultado el 14 de Agosto de 2020, en https://towardsdatascience.com/how-i-understood-what-features-to-consider-while-training-audio-files-eedfb6e9002b\n",
        "\n",
        "\n"
      ]
    }
  ]
}