def train_vae():
print("found {} files".format(len(wav_files)))
# encoder = keras.models.load_model("vae/encoder-test2.h")
# decoder = keras.models.load_model("vae/decoder-test2.h")
train_generator = NoteIsoSequence(train_wav_files,
sample_duration=sample_duration,
sample_rate=sample_rate,
n_fft=n_fft,
batch_size=batch_size,
epsilon=epsilon,
song_indices=song_indices,
instr_indices=instr_indices,
note_indices=note_indices)
test_generator = NoteIsoSequence(test_wav_files,
sample_duration=sample_duration,
sample_rate=sample_rate,
n_fft=n_fft,
batch_size=batch_size,
epsilon=epsilon,
song_indices=song_indices,
instr_indices=instr_indices,
note_indices=note_indices)
# tb = keras.callbacks.TensorBoard(histogram_freq=0, write_grads=True)
encoder, decoder, vae, my_vae_loss = get_vae()
vae.summary()
now = datetime.now()
log_dir = "logs/vae-" + now.strftime("%Y-%m-%d-%H:%M:%S") + "/"
callbacks = [
TensorBoardWrapper(test_generator,
log_dir=log_dir,
nb_steps=5,
histogram_freq=0,
batch_size=batch_size,
write_graph=False,
write_grads=True,
write_images=False)
]
vae.fit_generator(generator=train_generator,
validation_data=test_generator,
use_multiprocessing=use_multiprocessing,
workers=workers,
epochs=epochs,
steps_per_epoch=steps_per_epoch,
validation_steps=validation_steps,
callbacks=callbacks)
print("saving model...")
encoder.save("vae/encoder-test2.h")
decoder.save("vae/decoder-test2.h")
vae.save("vae/vae-test.h")
print("saved vae.")
},
'ae': {
'time': 0,
'statistical': {'m(mean)': 0, 'm(std)': 0, 'm(skew)': 0, 'm(kurtosis)': 0},
'clustering': {'ARI': 0},
'classification': {'m1d1': 0, 'm2d1': 0}
}
}
pr_dat_train = preprocessed_data.iloc[train_index]
pr_dat_test = preprocessed_data.iloc[test_index]
dat_train = data.iloc[train_index]
dat_test = data.iloc[test_index]
# vae gen
start = timer()
gen, enc = get_vae(pr_dat_train)
new_data = generate_data_vae(generator=gen,
encoder=enc,
org_data=pr_dat_train,
n=1000)
end = timer()
time_vae = end - start
results['vae']['time'] += time_vae
stats_eval = evaluation.statistical(pr_dat_train,
new_data,
numerical_cols=numerical_cols)
results['vae']['statistical']['m(mean)'] += stats_eval[0]
results['vae']['statistical']['m(std)'] += stats_eval[1]
results['vae']['statistical']['m(skew)'] += stats_eval[2]
results['vae']['statistical']['m(kurtosis)'] += stats_eval[3]
import model as model
import keras.utils as utils
import keras.datasets as datasets
import matplotlib.pyplot as plt
import numpy as np
import keras.backend as K
import utility
image_set = utility.get_MNIST()
vae_model = model.get_vae(image_set.shape[1:])
history = vae_model.fit(image_set, image_set, batch_size=128, epochs=10)
enco, deco = vae_model.layers
rand = np.random.randint(0, image_set.shape[0], (4))
for i in range(4):
realim = image_set[rand[i]]
predim = vae_model.predict(realim[np.newaxis, ...])
realim, predim = [np.reshape(x, (28, 28)) for x in [realim, predim]]
plt.subplot(4, 2, i * 2 + 1)
plt.imshow(realim, cmap='gray')
plt.subplot(4, 2, i * 2 + 2)
plt.imshow(predim, cmap='gray')
n = 4
randSamples = K.eval(K.random_normal((n * n, 10)))
predic_imag = deco.predict(randSamples)
for i in range(n * n):
plt.subplot(n, n, i + 1)