def training(autoencoder, encoder, augment, batch):
sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9,
nesterov=True) # Adam(lr=0.01)
autoencoder.compile(optimizer=sgd,
loss='binary_crossentropy',
metrics=['accuracy'])
x_train, x_test = dataset_loader.load_data_wrapper()
train_set, test_set, valid_set = dataset_loader.data_preprocessing(
x_train, x_test, augment)
early = EarlyStopping(monitor='val_loss',
min_delta=0,
patience=50,
mode='min',
verbose=1)
autoencoder.fit(train_set[0],
train_set[1],
epochs=500,
batch_size=batch,
shuffle=True,
validation_data=(valid_set[0], valid_set[1]),
callbacks=[early])
# TensorBoard(log_dir='C:/Users/213539359/Downloads/AlexNet-Tensorflow-master/logs/'),
# autoencoder.save('C:/Users/213539359/Downloads/AlexNet-Tensorflow-master/model/autoencoder_3.h5')
encoded_img = encoder.predict(test_set[0])
decoded_imgs = autoencoder.predict(test_set[0])
results = autoencoder.evaluate(test_set[0], test_set[1], batch_size=1)
print("TEST LOSS, TEST ACC: ", results)
return x_test, decoded_imgs, encoded_img
import dataset_loader
training_data, validation_data, test_data = dataset_loader.load_data_wrapper()
import neuralnet
x = int(input("Enter the no. of neurons in the hidden layer: "))
net = neuralnet.NeuralNet([784, x, 10])
e = int(input("Enter the no. of epochs: "))
s = int(input("Enter the size of each batch: "))
l = int(input("Enter the learning rate: "))
net.train(training_set=training_data,
no_of_epochs=e,
size_of_batch=s,
learning_rate=l,
test_set=test_data)