Training_Model.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Training_Model.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "o_swHddOGP_0",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "54281888-11fc-463e-b95e-4ce9ad4ead64"
      },
      "source": [
        "\n",
        "\"\"\"\n",
        "Created on Thu Jul 16 12:30:12 2020\n",
        "\n",
        "@author: Anshdeep\n",
        "\n",
        "The following program is a image classification training model based on the CIFAR-10 dataset which is a collection \n",
        "of images of 10 different classes like cars, birds, dogs, horses, ships, trucks, etc. and is entierly based on training\n",
        "the model and plots the test and train accuracies and losses and saves the data into CIFAR10_.h5 file which can be imported \n",
        "and tested as done in Testing_Model.ipynb file. \n",
        "\"\"\"\n",
        " \n",
        "import pandas as pd\n",
        "from matplotlib import pyplot\n",
        "from keras.datasets import cifar10\n",
        "from keras.utils import to_categorical\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Dense\n",
        "from keras.layers import BatchNormalization\n",
        "from keras.layers import Dropout\n",
        "from keras.layers import MaxPooling2D\n",
        "from keras.layers import Conv2D\n",
        "from keras.layers import Flatten\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "\n",
        "# This function loads the images to x_test , x_train , y_test and y_train and normailzes the x_test and x_train and categorize y_test and y_train \n",
        "def loading_dataset():\n",
        "    (x_train, y_train), (x_test, y_test) = cifar10.load_data()\n",
        "    y1_train = to_categorical(y_train)\n",
        "    y1_test = to_categorical(y_test)\n",
        "    \n",
        "    x_test  = x_test/255.0\n",
        "    x_train  =  x_train/255.0\n",
        "    \n",
        "    return x_test , x_train , y1_test , y1_train , y_test , y_train\n",
        "\n",
        "# Defining the model to train data \n",
        "def training_model():\n",
        "    model= Sequential()\n",
        "    # Adding a 2D convolutional layer with 32 filters, kernel size of 3x3 and kernel initilizer as he_uniform\n",
        "    # Dropouts help reduce overfitting and BatchNormalization to enhance the learning quality in each layer\n",
        "    model.add(Conv2D(32 , ( 3,3 ) , input_shape = ( 32,32,3 ) , activation = 'relu' , kernel_initializer = 'he_uniform' , padding = 'same' ) ) # input layer\n",
        "    model.add(Conv2D(32 , ( 3,3 ) , activation = 'relu' , kernel_initializer = 'he_uniform' , padding = 'same' ) ) # Following are the hidden layers of the model\n",
        "    model.add(BatchNormalization())\n",
        "    model.add(MaxPooling2D( ( 2,2 ) ) )\n",
        "    model.add(Dropout(0.2))\n",
        "    \n",
        "    model.add(Conv2D(64 , ( 3,3 ) , activation = 'relu' , kernel_initializer = 'he_uniform' , padding = 'same' ) )\n",
        "    model.add(BatchNormalization())\n",
        "    model.add(Conv2D(64 , ( 3,3 ) , activation = 'relu' , kernel_initializer = 'he_uniform' , padding = 'same' ) )\n",
        "    model.add(BatchNormalization())\n",
        "    model.add(MaxPooling2D( ( 2,2 ) ) )\n",
        "    model.add(Dropout(0.3))\n",
        "    \n",
        "    model.add(Conv2D(128 , ( 3,3 ) , activation = 'relu' , kernel_initializer = 'he_uniform' , padding = 'same' ) )\n",
        "    model.add(BatchNormalization())\n",
        "    model.add(Conv2D(128 , ( 3,3 ) , activation = 'relu' , kernel_initializer = 'he_uniform' , padding = 'same' ) )\n",
        "    model.add(BatchNormalization())\n",
        "    model.add(MaxPooling2D( ( 2,2 ) ) )\n",
        "    model.add(Dropout(0.5))\n",
        "    \n",
        "    model.add(Conv2D(256 , ( 3,3 ) , activation = 'relu' , kernel_initializer = 'he_uniform' , padding = 'same' ) )\n",
        "    model.add(BatchNormalization())\n",
        "    model.add(Conv2D(256 , ( 3,3 ) , activation = 'relu' , kernel_initializer = 'he_uniform' , padding = 'same' ) )\n",
        "    model.add(BatchNormalization())\n",
        "    model.add(MaxPooling2D( ( 2,2 ) ) )\n",
        "    model.add(Dropout(0.5))\n",
        "\n",
        "    model.add(Conv2D(512 , ( 3,3 ) , activation = 'relu' , kernel_initializer = 'he_uniform' , padding = 'same' ) )\n",
        "    model.add(BatchNormalization())\n",
        "    model.add(Conv2D(512 , ( 3,3 ) , activation = 'relu' , kernel_initializer = 'he_uniform' , padding = 'same' ) )\n",
        "    model.add(BatchNormalization())\n",
        "    model.add(MaxPooling2D( ( 2,2 ) ) )\n",
        "    model.add(Dropout(0.5))\n",
        "    \n",
        "    model.add(Flatten()) \n",
        "    model.add(Dense(256, activation = 'relu' , kernel_initializer = 'he_uniform'))\n",
        "    model.add(BatchNormalization())\n",
        "    model.add(Dropout(0.5))\n",
        "    model.add(Dense(10 , activation = 'softmax')) # Output Layer\n",
        "    \n",
        "    model.compile(optimizer = 'adam' , loss = 'categorical_crossentropy' , metrics=[\"accuracy\"])\n",
        "    \n",
        "    return model\n",
        "\n",
        "# Loads the data and trains it and evaluates it\n",
        "def test_run():\n",
        "    x_test, x_train , y1_test , y1_train , y_test , y_train = loading_dataset() # Storing the values returned by Loading_dataset()   \n",
        "    \n",
        "    model = training_model() # Storing the traning model \n",
        "    \n",
        "    datagen = ImageDataGenerator(width_shift_range=0.2, height_shift_range=0.2, horizontal_flip=True) # Minimizing overfitting\n",
        "    it_train = datagen.flow(x_train, y1_train, batch_size=64)\n",
        "    # Training the Model\n",
        "    saved_model = model.fit_generator( it_train , epochs = 150 , validation_data= ( x_test , y1_test ))\n",
        "    # Accuracy and Loss plot for train and test data\n",
        "    eval1 = pd.DataFrame(model.history.history)\n",
        "    eval1.plot()\n",
        "    print(eval1)\n",
        "\n",
        "    metrics = pd.DataFrame(saved_model.history)\n",
        "    metrics[['accuracy' , 'val_accuracy']].plot()\n",
        "\n",
        "    return model\n",
        "\n",
        "x  = test_run()\n",
        "x.save('CIFAR10_.h5') # Saves the model into as CIFAR10_.h5 file\n",
        "\n"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/150\n",
            "782/782 [==============================] - 50s 64ms/step - loss: 2.1118 - accuracy: 0.2773 - val_loss: 1.6922 - val_accuracy: 0.3776\n",
            "Epoch 2/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 1.5431 - accuracy: 0.4346 - val_loss: 1.7561 - val_accuracy: 0.4371\n",
            "Epoch 3/150\n",
            "782/782 [==============================] - 49s 63ms/step - loss: 1.3305 - accuracy: 0.5199 - val_loss: 1.3284 - val_accuracy: 0.5497\n",
            "Epoch 4/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 1.1782 - accuracy: 0.5835 - val_loss: 1.0844 - val_accuracy: 0.6265\n",
            "Epoch 5/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 1.0574 - accuracy: 0.6317 - val_loss: 0.8909 - val_accuracy: 0.6896\n",
            "Epoch 6/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.9768 - accuracy: 0.6620 - val_loss: 0.8398 - val_accuracy: 0.7180\n",
            "Epoch 7/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.9076 - accuracy: 0.6895 - val_loss: 0.9048 - val_accuracy: 0.6894\n",
            "Epoch 8/150\n",
            "782/782 [==============================] - 47s 61ms/step - loss: 0.8652 - accuracy: 0.7074 - val_loss: 0.7435 - val_accuracy: 0.7404\n",
            "Epoch 9/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.8246 - accuracy: 0.7212 - val_loss: 0.8029 - val_accuracy: 0.7278\n",
            "Epoch 10/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.7913 - accuracy: 0.7318 - val_loss: 0.7762 - val_accuracy: 0.7355\n",
            "Epoch 11/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.7568 - accuracy: 0.7432 - val_loss: 0.6056 - val_accuracy: 0.7960\n",
            "Epoch 12/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.7326 - accuracy: 0.7517 - val_loss: 0.6412 - val_accuracy: 0.7811\n",
            "Epoch 13/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.7125 - accuracy: 0.7623 - val_loss: 0.6495 - val_accuracy: 0.7797\n",
            "Epoch 14/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.6913 - accuracy: 0.7685 - val_loss: 0.5862 - val_accuracy: 0.8056\n",
            "Epoch 15/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.6730 - accuracy: 0.7761 - val_loss: 0.7453 - val_accuracy: 0.7518\n",
            "Epoch 16/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.6579 - accuracy: 0.7795 - val_loss: 0.5619 - val_accuracy: 0.8062\n",
            "Epoch 17/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.6435 - accuracy: 0.7850 - val_loss: 0.5438 - val_accuracy: 0.8181\n",
            "Epoch 18/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.6226 - accuracy: 0.7913 - val_loss: 0.5022 - val_accuracy: 0.8278\n",
            "Epoch 19/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.6186 - accuracy: 0.7930 - val_loss: 0.5004 - val_accuracy: 0.8299\n",
            "Epoch 20/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.6051 - accuracy: 0.7975 - val_loss: 0.5365 - val_accuracy: 0.8199\n",
            "Epoch 21/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.5915 - accuracy: 0.8017 - val_loss: 0.6319 - val_accuracy: 0.7884\n",
            "Epoch 22/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.5745 - accuracy: 0.8078 - val_loss: 0.5496 - val_accuracy: 0.8187\n",
            "Epoch 23/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.5748 - accuracy: 0.8086 - val_loss: 0.5535 - val_accuracy: 0.8171\n",
            "Epoch 24/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.5649 - accuracy: 0.8099 - val_loss: 0.4808 - val_accuracy: 0.8417\n",
            "Epoch 25/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.5528 - accuracy: 0.8136 - val_loss: 0.4883 - val_accuracy: 0.8382\n",
            "Epoch 26/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.5470 - accuracy: 0.8167 - val_loss: 0.4728 - val_accuracy: 0.8438\n",
            "Epoch 27/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.5338 - accuracy: 0.8207 - val_loss: 0.5313 - val_accuracy: 0.8255\n",
            "Epoch 28/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.5301 - accuracy: 0.8215 - val_loss: 0.6727 - val_accuracy: 0.7890\n",
            "Epoch 29/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.5277 - accuracy: 0.8237 - val_loss: 0.4832 - val_accuracy: 0.8386\n",
            "Epoch 30/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.5150 - accuracy: 0.8263 - val_loss: 0.4825 - val_accuracy: 0.8378\n",
            "Epoch 31/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.5081 - accuracy: 0.8296 - val_loss: 0.4198 - val_accuracy: 0.8595\n",
            "Epoch 32/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.5059 - accuracy: 0.8324 - val_loss: 0.4832 - val_accuracy: 0.8408\n",
            "Epoch 33/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4994 - accuracy: 0.8326 - val_loss: 0.5359 - val_accuracy: 0.8240\n",
            "Epoch 34/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4979 - accuracy: 0.8319 - val_loss: 0.5551 - val_accuracy: 0.8144\n",
            "Epoch 35/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4919 - accuracy: 0.8336 - val_loss: 0.4221 - val_accuracy: 0.8571\n",
            "Epoch 36/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4860 - accuracy: 0.8371 - val_loss: 0.4324 - val_accuracy: 0.8576\n",
            "Epoch 37/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4765 - accuracy: 0.8380 - val_loss: 0.4497 - val_accuracy: 0.8518\n",
            "Epoch 38/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4755 - accuracy: 0.8415 - val_loss: 0.4352 - val_accuracy: 0.8531\n",
            "Epoch 39/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4716 - accuracy: 0.8406 - val_loss: 0.4167 - val_accuracy: 0.8620\n",
            "Epoch 40/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4675 - accuracy: 0.8430 - val_loss: 0.4072 - val_accuracy: 0.8661\n",
            "Epoch 41/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4661 - accuracy: 0.8427 - val_loss: 0.4322 - val_accuracy: 0.8589\n",
            "Epoch 42/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.4611 - accuracy: 0.8468 - val_loss: 0.4077 - val_accuracy: 0.8650\n",
            "Epoch 43/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.4508 - accuracy: 0.8485 - val_loss: 0.3923 - val_accuracy: 0.8691\n",
            "Epoch 44/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4524 - accuracy: 0.8475 - val_loss: 0.3826 - val_accuracy: 0.8711\n",
            "Epoch 45/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4478 - accuracy: 0.8502 - val_loss: 0.3998 - val_accuracy: 0.8683\n",
            "Epoch 46/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4475 - accuracy: 0.8497 - val_loss: 0.4510 - val_accuracy: 0.8515\n",
            "Epoch 47/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4394 - accuracy: 0.8534 - val_loss: 0.3767 - val_accuracy: 0.8761\n",
            "Epoch 48/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4409 - accuracy: 0.8513 - val_loss: 0.4031 - val_accuracy: 0.8654\n",
            "Epoch 49/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4367 - accuracy: 0.8536 - val_loss: 0.4088 - val_accuracy: 0.8642\n",
            "Epoch 50/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4347 - accuracy: 0.8535 - val_loss: 0.4160 - val_accuracy: 0.8629\n",
            "Epoch 51/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4273 - accuracy: 0.8560 - val_loss: 0.4218 - val_accuracy: 0.8593\n",
            "Epoch 52/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4267 - accuracy: 0.8567 - val_loss: 0.4004 - val_accuracy: 0.8668\n",
            "Epoch 53/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4295 - accuracy: 0.8570 - val_loss: 0.3644 - val_accuracy: 0.8769\n",
            "Epoch 54/150\n",
            "782/782 [==============================] - 47s 59ms/step - loss: 0.4157 - accuracy: 0.8600 - val_loss: 0.3917 - val_accuracy: 0.8693\n",
            "Epoch 55/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4205 - accuracy: 0.8598 - val_loss: 0.3644 - val_accuracy: 0.8761\n",
            "Epoch 56/150\n",
            "782/782 [==============================] - 47s 59ms/step - loss: 0.4155 - accuracy: 0.8608 - val_loss: 0.4176 - val_accuracy: 0.8615\n",
            "Epoch 57/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4126 - accuracy: 0.8611 - val_loss: 0.4141 - val_accuracy: 0.8657\n",
            "Epoch 58/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4097 - accuracy: 0.8628 - val_loss: 0.4026 - val_accuracy: 0.8681\n",
            "Epoch 59/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4036 - accuracy: 0.8644 - val_loss: 0.3625 - val_accuracy: 0.8836\n",
            "Epoch 60/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4106 - accuracy: 0.8619 - val_loss: 0.3905 - val_accuracy: 0.8735\n",
            "Epoch 61/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.4044 - accuracy: 0.8635 - val_loss: 0.4575 - val_accuracy: 0.8503\n",
            "Epoch 62/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3952 - accuracy: 0.8659 - val_loss: 0.4610 - val_accuracy: 0.8469\n",
            "Epoch 63/150\n",
            "782/782 [==============================] - 47s 59ms/step - loss: 0.3964 - accuracy: 0.8659 - val_loss: 0.3795 - val_accuracy: 0.8730\n",
            "Epoch 64/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3939 - accuracy: 0.8673 - val_loss: 0.3924 - val_accuracy: 0.8711\n",
            "Epoch 65/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3905 - accuracy: 0.8680 - val_loss: 0.4513 - val_accuracy: 0.8541\n",
            "Epoch 66/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3904 - accuracy: 0.8686 - val_loss: 0.3658 - val_accuracy: 0.8792\n",
            "Epoch 67/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3877 - accuracy: 0.8687 - val_loss: 0.3452 - val_accuracy: 0.8876\n",
            "Epoch 68/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.3966 - accuracy: 0.8670 - val_loss: 0.3855 - val_accuracy: 0.8718\n",
            "Epoch 69/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3793 - accuracy: 0.8726 - val_loss: 0.4182 - val_accuracy: 0.8613\n",
            "Epoch 70/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3846 - accuracy: 0.8698 - val_loss: 0.3897 - val_accuracy: 0.8701\n",
            "Epoch 71/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3796 - accuracy: 0.8708 - val_loss: 0.3712 - val_accuracy: 0.8795\n",
            "Epoch 72/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3790 - accuracy: 0.8717 - val_loss: 0.3628 - val_accuracy: 0.8789\n",
            "Epoch 73/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3795 - accuracy: 0.8721 - val_loss: 0.3426 - val_accuracy: 0.8879\n",
            "Epoch 74/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3803 - accuracy: 0.8711 - val_loss: 0.3633 - val_accuracy: 0.8796\n",
            "Epoch 75/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3693 - accuracy: 0.8757 - val_loss: 0.3629 - val_accuracy: 0.8825\n",
            "Epoch 76/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3784 - accuracy: 0.8721 - val_loss: 0.3689 - val_accuracy: 0.8785\n",
            "Epoch 77/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3656 - accuracy: 0.8759 - val_loss: 0.3732 - val_accuracy: 0.8794\n",
            "Epoch 78/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3693 - accuracy: 0.8748 - val_loss: 0.3849 - val_accuracy: 0.8704\n",
            "Epoch 79/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3661 - accuracy: 0.8758 - val_loss: 0.3349 - val_accuracy: 0.8882\n",
            "Epoch 80/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3671 - accuracy: 0.8760 - val_loss: 0.4140 - val_accuracy: 0.8648\n",
            "Epoch 81/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3626 - accuracy: 0.8773 - val_loss: 0.3638 - val_accuracy: 0.8807\n",
            "Epoch 82/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3649 - accuracy: 0.8764 - val_loss: 0.3548 - val_accuracy: 0.8847\n",
            "Epoch 83/150\n",
            "782/782 [==============================] - 47s 59ms/step - loss: 0.3615 - accuracy: 0.8774 - val_loss: 0.3423 - val_accuracy: 0.8863\n",
            "Epoch 84/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.3577 - accuracy: 0.8793 - val_loss: 0.3457 - val_accuracy: 0.8842\n",
            "Epoch 85/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3573 - accuracy: 0.8774 - val_loss: 0.3646 - val_accuracy: 0.8834\n",
            "Epoch 86/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3607 - accuracy: 0.8762 - val_loss: 0.3544 - val_accuracy: 0.8835\n",
            "Epoch 87/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3574 - accuracy: 0.8794 - val_loss: 0.3391 - val_accuracy: 0.8862\n",
            "Epoch 88/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3542 - accuracy: 0.8794 - val_loss: 0.3179 - val_accuracy: 0.8954\n",
            "Epoch 89/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3511 - accuracy: 0.8819 - val_loss: 0.3589 - val_accuracy: 0.8845\n",
            "Epoch 90/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3521 - accuracy: 0.8806 - val_loss: 0.3206 - val_accuracy: 0.8930\n",
            "Epoch 91/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3454 - accuracy: 0.8827 - val_loss: 0.3554 - val_accuracy: 0.8847\n",
            "Epoch 92/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.3534 - accuracy: 0.8817 - val_loss: 0.3253 - val_accuracy: 0.8918\n",
            "Epoch 93/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3503 - accuracy: 0.8802 - val_loss: 0.3566 - val_accuracy: 0.8812\n",
            "Epoch 94/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3413 - accuracy: 0.8838 - val_loss: 0.3429 - val_accuracy: 0.8885\n",
            "Epoch 95/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3421 - accuracy: 0.8856 - val_loss: 0.3814 - val_accuracy: 0.8734\n",
            "Epoch 96/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3450 - accuracy: 0.8834 - val_loss: 0.3516 - val_accuracy: 0.8838\n",
            "Epoch 97/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3393 - accuracy: 0.8845 - val_loss: 0.3325 - val_accuracy: 0.8898\n",
            "Epoch 98/150\n",
            "782/782 [==============================] - 47s 59ms/step - loss: 0.3411 - accuracy: 0.8843 - val_loss: 0.3302 - val_accuracy: 0.8882\n",
            "Epoch 99/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.3450 - accuracy: 0.8838 - val_loss: 0.3463 - val_accuracy: 0.8845\n",
            "Epoch 100/150\n",
            "782/782 [==============================] - 47s 59ms/step - loss: 0.3402 - accuracy: 0.8855 - val_loss: 0.3227 - val_accuracy: 0.8928\n",
            "Epoch 101/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3386 - accuracy: 0.8847 - val_loss: 0.3355 - val_accuracy: 0.8888\n",
            "Epoch 102/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.3341 - accuracy: 0.8868 - val_loss: 0.3851 - val_accuracy: 0.8748\n",
            "Epoch 103/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3311 - accuracy: 0.8876 - val_loss: 0.3195 - val_accuracy: 0.8952\n",
            "Epoch 104/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3310 - accuracy: 0.8873 - val_loss: 0.3744 - val_accuracy: 0.8778\n",
            "Epoch 105/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3314 - accuracy: 0.8865 - val_loss: 0.3646 - val_accuracy: 0.8841\n",
            "Epoch 106/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3292 - accuracy: 0.8881 - val_loss: 0.3143 - val_accuracy: 0.8986\n",
            "Epoch 107/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3287 - accuracy: 0.8871 - val_loss: 0.3354 - val_accuracy: 0.8894\n",
            "Epoch 108/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3319 - accuracy: 0.8890 - val_loss: 0.3504 - val_accuracy: 0.8857\n",
            "Epoch 109/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3282 - accuracy: 0.8876 - val_loss: 0.3261 - val_accuracy: 0.8900\n",
            "Epoch 110/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3274 - accuracy: 0.8896 - val_loss: 0.3220 - val_accuracy: 0.8970\n",
            "Epoch 111/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3215 - accuracy: 0.8906 - val_loss: 0.3408 - val_accuracy: 0.8868\n",
            "Epoch 112/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3222 - accuracy: 0.8916 - val_loss: 0.3703 - val_accuracy: 0.8803\n",
            "Epoch 113/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3187 - accuracy: 0.8914 - val_loss: 0.3186 - val_accuracy: 0.8923\n",
            "Epoch 114/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3204 - accuracy: 0.8913 - val_loss: 0.3311 - val_accuracy: 0.8921\n",
            "Epoch 115/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3237 - accuracy: 0.8911 - val_loss: 0.3064 - val_accuracy: 0.8977\n",
            "Epoch 116/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3219 - accuracy: 0.8920 - val_loss: 0.3345 - val_accuracy: 0.8896\n",
            "Epoch 117/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3206 - accuracy: 0.8920 - val_loss: 0.3116 - val_accuracy: 0.8950\n",
            "Epoch 118/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3173 - accuracy: 0.8919 - val_loss: 0.3190 - val_accuracy: 0.8961\n",
            "Epoch 119/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3143 - accuracy: 0.8930 - val_loss: 0.3068 - val_accuracy: 0.8997\n",
            "Epoch 120/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3203 - accuracy: 0.8898 - val_loss: 0.3205 - val_accuracy: 0.8925\n",
            "Epoch 121/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.3188 - accuracy: 0.8918 - val_loss: 0.3318 - val_accuracy: 0.8899\n",
            "Epoch 122/150\n",
            "782/782 [==============================] - 47s 59ms/step - loss: 0.3129 - accuracy: 0.8941 - val_loss: 0.2929 - val_accuracy: 0.9045\n",
            "Epoch 123/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.3101 - accuracy: 0.8950 - val_loss: 0.3246 - val_accuracy: 0.8949\n",
            "Epoch 124/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.3123 - accuracy: 0.8931 - val_loss: 0.3297 - val_accuracy: 0.8916\n",
            "Epoch 125/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3107 - accuracy: 0.8966 - val_loss: 0.3330 - val_accuracy: 0.8897\n",
            "Epoch 126/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.3026 - accuracy: 0.8948 - val_loss: 0.3016 - val_accuracy: 0.9019\n",
            "Epoch 127/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3047 - accuracy: 0.8964 - val_loss: 0.3565 - val_accuracy: 0.8855\n",
            "Epoch 128/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3125 - accuracy: 0.8951 - val_loss: 0.3481 - val_accuracy: 0.8884\n",
            "Epoch 129/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3043 - accuracy: 0.8974 - val_loss: 0.3568 - val_accuracy: 0.8843\n",
            "Epoch 130/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3001 - accuracy: 0.8976 - val_loss: 0.3238 - val_accuracy: 0.8907\n",
            "Epoch 131/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3012 - accuracy: 0.8964 - val_loss: 0.3487 - val_accuracy: 0.8867\n",
            "Epoch 132/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3067 - accuracy: 0.8963 - val_loss: 0.3160 - val_accuracy: 0.8945\n",
            "Epoch 133/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.3012 - accuracy: 0.8967 - val_loss: 0.2953 - val_accuracy: 0.9014\n",
            "Epoch 134/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3032 - accuracy: 0.8971 - val_loss: 0.3265 - val_accuracy: 0.8929\n",
            "Epoch 135/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.2986 - accuracy: 0.8973 - val_loss: 0.3332 - val_accuracy: 0.8922\n",
            "Epoch 136/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.3006 - accuracy: 0.8982 - val_loss: 0.3133 - val_accuracy: 0.8990\n",
            "Epoch 137/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.2987 - accuracy: 0.8981 - val_loss: 0.3175 - val_accuracy: 0.8935\n",
            "Epoch 138/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3004 - accuracy: 0.8986 - val_loss: 0.3092 - val_accuracy: 0.8977\n",
            "Epoch 139/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.3016 - accuracy: 0.8985 - val_loss: 0.3141 - val_accuracy: 0.8954\n",
            "Epoch 140/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.2931 - accuracy: 0.8996 - val_loss: 0.3270 - val_accuracy: 0.8931\n",
            "Epoch 141/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.2960 - accuracy: 0.8995 - val_loss: 0.3096 - val_accuracy: 0.8972\n",
            "Epoch 142/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.2919 - accuracy: 0.9003 - val_loss: 0.3155 - val_accuracy: 0.8999\n",
            "Epoch 143/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.2984 - accuracy: 0.8986 - val_loss: 0.3123 - val_accuracy: 0.8988\n",
            "Epoch 144/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.2944 - accuracy: 0.8999 - val_loss: 0.3410 - val_accuracy: 0.8890\n",
            "Epoch 145/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.2908 - accuracy: 0.9012 - val_loss: 0.3260 - val_accuracy: 0.8950\n",
            "Epoch 146/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.2965 - accuracy: 0.8995 - val_loss: 0.3181 - val_accuracy: 0.8965\n",
            "Epoch 147/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.2948 - accuracy: 0.8987 - val_loss: 0.3534 - val_accuracy: 0.8867\n",
            "Epoch 148/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.2902 - accuracy: 0.9010 - val_loss: 0.3239 - val_accuracy: 0.8945\n",
            "Epoch 149/150\n",
            "782/782 [==============================] - 46s 59ms/step - loss: 0.2879 - accuracy: 0.9023 - val_loss: 0.3184 - val_accuracy: 0.8978\n",
            "Epoch 150/150\n",
            "782/782 [==============================] - 47s 60ms/step - loss: 0.2914 - accuracy: 0.8996 - val_loss: 0.3083 - val_accuracy: 0.9012\n",
            "     val_loss  val_accuracy      loss  accuracy\n",
            "0    1.692187        0.3776  2.112719   0.27730\n",
            "1    1.756055        0.4371  1.543670   0.43464\n",
            "2    1.328445        0.5497  1.330783   0.51988\n",
            "3    1.084422        0.6265  1.177890   0.58350\n",
            "4    0.890948        0.6896  1.057610   0.63166\n",
            "..        ...           ...       ...       ...\n",
            "145  0.318131        0.8965  0.295906   0.89950\n",
            "146  0.353371        0.8867  0.294674   0.89872\n",
            "147  0.323933        0.8945  0.290099   0.90102\n",
            "148  0.318441        0.8978  0.287394   0.90232\n",
            "149  0.308316        0.9012  0.291114   0.89958\n",
            "\n",
            "[150 rows x 4 columns]\n"
          ],
          "name": "stdout"
        },
        {
          "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"
          }
        },
        {
          "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": "code",
      "metadata": {
        "colab_type": "code",
        "id": "h1AnV13C5FJ6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 470
        },
        "outputId": "720c6d9e-7341-4521-c8a7-d89d4332b0c0"
      },
      "source": [
        "import numpy as np\n",
        "from sklearn.metrics import classification_report, confusion_matrix\n",
        "(p_x_train , p_y_train) , (p_x_test , p_y_test) = cifar10.load_data() \n",
        "p_x_test = p_x_test/255.0\n",
        "p_x_train = p_x_train/255.0\n",
        "predictions = np.argmax(x.predict(p_x_test), axis=-1)\n",
        "\n",
        "print(classification_report(p_y_test ,predictions))\n",
        "print(confusion_matrix(p_y_test ,predictions))"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.92      0.93      0.92      1000\n",
            "           1       0.96      0.96      0.96      1000\n",
            "           2       0.91      0.84      0.87      1000\n",
            "           3       0.84      0.75      0.79      1000\n",
            "           4       0.87      0.93      0.90      1000\n",
            "           5       0.88      0.81      0.84      1000\n",
            "           6       0.83      0.97      0.89      1000\n",
            "           7       0.93      0.92      0.93      1000\n",
            "           8       0.94      0.95      0.95      1000\n",
            "           9       0.95      0.96      0.95      1000\n",
            "\n",
            "    accuracy                           0.90     10000\n",
            "   macro avg       0.90      0.90      0.90     10000\n",
            "weighted avg       0.90      0.90      0.90     10000\n",
            "\n",
            "[[927   4  11   7   8   0   2   5  27   9]\n",
            " [  3 961   0   0   0   0   2   0   6  28]\n",
            " [ 28   1 837  16  27  26  48  10   5   2]\n",
            " [  9   2  34 749  42  69  71  14   8   2]\n",
            " [  2   1  11   6 932   6  26  15   1   0]\n",
            " [  2   2  12  90  28 807  35  22   1   1]\n",
            " [  4   1   8  12   2   1 971   0   1   0]\n",
            " [  3   0   6  14  28  10  12 922   2   3]\n",
            " [ 26   6   1   1   1   0   5   0 950  10]\n",
            " [  7  23   1   2   1   0   2   0   8 956]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nu6GHUyt4xL-",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "x.save('CIFAR10_.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EonH8epV8mYi",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}