def build(self, optimizer='Adam'):
# build encoder
enc_input = Input(shape=self.input_shape)
enc_nn = NN(self.encoder_network)
enc_out = enc_nn.build(enc_input)
code_mean = Dense(self.code_size)(enc_out)
code_var = Dense(self.code_size)(enc_out)
code = Add()([
code_mean,
Multiply()(
[Lambda(K.exp)(code_var),
Lambda(random_normal)(code_var)])
])
enc_obj = Subtract()([
Add()([Lambda(K.square)(code_mean),
Lambda(K.exp)(code_var)]), code_var
])
self.encoder = Model(enc_input, [code, enc_obj], name='encoder')
# build decoder
dec_input = Input(shape=[self.code_size])
dec_nn = NN(self.decoder_network)
dec_out = dec_nn.build(dec_input)
self.decoder = Model(dec_input, dec_out, name='decoder')
# compose VAE
real_img = Input(shape=self.input_shape)
z, z_obj = self.encoder(real_img)
reconstruct_img = self.decoder(z)
self.vae = Model(inputs=[real_img], outputs=[reconstruct_img, z_obj])
self.vae.compile(optimizer=optimizer,
loss=[self.reconstruct_loss, self.encoding_loss],
loss_weights=self.loss_weights)
for experiment in experiments:
augment, batch_size, keep_prob, filters, ksize, config = experiment
start_time = time.time()
network = NN(epochs=5, batch_size=batch_size, learning_rate=0.001)
features = np.concatenate([X_train, aug_X
]) if augment == 'True' else X_train
labels = np.concatenate([y_train, aug_y]) if augment == 'True' else y_train
network.add_train_data(features, labels)
network.add_test_data(X_test, y_test)
network.add_validation_data(X_valid, y_valid)
network.add_configuration(create_network_config(config, filters, ksize),
input_size=input_size)
network.build(num_labels=num_labels)
print(
"Training model with hyperparameters: augmented: {}, batch_size: {}, keep_prob: {}, filters: {}, ksize: {}, config: {}"
.format(augment, batch_size, keep_prob, filters, ksize,
network.get_string()))
validation_accuracy = network.train(keep_prob=keep_prob)
end_time = time.time()
stat_values = [
end_time - start_time, validation_accuracy, augment, batch_size,
keep_prob, filters, ksize,
network.get_string()
]
stat_entry = pd.Series(stat_values, index=stat_labels)
stats = stats.append(stat_entry, ignore_index=True)
import numpy
from nn import NN
import pandas as pd
model = NN(10)
data = pd.read_csv('xor.csv', header=None).to_numpy()
xs = data[:, :2]
ys = data[:, 2:]
model.build(2)
for i in range(len(data)):
print(model.backward(xs[i], ys[i]))