model_arch_python

#This is the current model in Keras/python as of 8/28/21. This is for version 1.0.7 of vocal_numeric

import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, InputLayer, Dropout, Conv1D, Conv2D, Flatten, Reshape, MaxPooling1D, MaxPooling2D, BatchNormalization, TimeDistributed
from tensorflow.keras.optimizers import Adam
sys.path.append('./resources/libraries')
import ei_tensorflow.training

# model architecture
model = Sequential()
model.add(Reshape((int(input_length / 40), 40), input_shape=(input_length, )))
model.add(Conv1D(16, kernel_size=3, activation='relu', padding='same'))
model.add(MaxPooling1D(pool_size=2, strides=2, padding='same'))
model.add(Conv1D(32, kernel_size=3, activation='relu', padding='same'))
model.add(MaxPooling1D(pool_size=2, strides=2, padding='same'))
model.add(Conv1D(64, kernel_size=3, activation='relu', padding='same'))
model.add(MaxPooling1D(pool_size=2, strides=2, padding='same'))
model.add(Flatten())
model.add(Dropout(0.25))
model.add(Dense(64, activation='relu',
    activity_regularizer=tf.keras.regularizers.l1(0.00001)))
model.add(Dropout(0.25))
model.add(Dense(classes, activation='softmax', name='y_pred'))

# this controls the learning rate
opt = Adam(lr=0.005, beta_1=0.9, beta_2=0.999)
# this controls the batch size, or you can manipulate the tf.data.Dataset objects yourself
BATCH_SIZE = 32
train_dataset = train_dataset.batch(BATCH_SIZE, drop_remainder=False)
validation_dataset = validation_dataset.batch(BATCH_SIZE, drop_remainder=False)
callbacks.append(BatchLoggerCallback(BATCH_SIZE, train_sample_count))

# train the neural network
model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy'])
model.fit(train_dataset, epochs=100, validation_data=validation_dataset, verbose=2, callbacks=callbacks)