Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

acgan: Add batch normalization to the Generator, etc #8616

Merged
merged 2 commits into from
Jan 19, 2018
Merged

acgan: Add batch normalization to the Generator, etc #8616

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
9fe4466
Add batch normalization to the Generator
ozabluda Nov 28, 2017
da6fb93
Add name of a paper
ozabluda Nov 28, 2017
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 9 additions & 4 deletions examples/mnist_acgan.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,6 +33,7 @@
from keras.datasets import mnist
from keras import layers
from keras.layers import Input, Dense, Reshape, Flatten, Embedding, Dropout
from keras.layers import BatchNormalization
from keras.layers.advanced_activations import LeakyReLU
from keras.layers.convolutional import Conv2DTranspose, Conv2D
from keras.models import Sequential, Model
Expand All @@ -56,11 +57,13 @@ def build_generator(latent_size):
cnn.add(Conv2DTranspose(192, 5, strides=1, padding='valid',
activation='relu',
kernel_initializer='glorot_normal'))
cnn.add(BatchNormalization())

# upsample to (14, 14, ...)
cnn.add(Conv2DTranspose(96, 5, strides=2, padding='same',
activation='relu',
kernel_initializer='glorot_normal'))
cnn.add(BatchNormalization())

# upsample to (28, 28, ...)
cnn.add(Conv2DTranspose(1, 5, strides=2, padding='same',
Expand Down Expand Up @@ -124,7 +127,7 @@ def build_discriminator():
if __name__ == '__main__':

# batch and latent size taken from the paper
epochs = 50
epochs = 100
batch_size = 100
latent_size = 100

Expand Down Expand Up @@ -215,8 +218,10 @@ def build_discriminator():

x = np.concatenate((image_batch, generated_images))

# use soft real/fake labels
soft_zero, soft_one = 0.1, 0.9
# use one-sided soft real/fake labels
# Salimans et al., 2016
# https://arxiv.org/pdf/1606.03498.pdf (Section 3.4)
soft_zero, soft_one = 0, 0.95
y = np.array([soft_one] * batch_size + [soft_zero] * batch_size)
aux_y = np.concatenate((label_batch, sampled_labels), axis=0)

Expand Down Expand Up @@ -286,7 +291,7 @@ def build_discriminator():
'component', *discriminator.metrics_names))
print('-' * 65)

ROW_FMT = '{0:<22s} | {1:<4.2f} | {2:<15.2f} | {3:<5.2f}'
ROW_FMT = '{0:<22s} | {1:<4.2f} | {2:<15.4f} | {3:<5.4f}'
print(ROW_FMT.format('generator (train)',
*train_history['generator'][-1]))
print(ROW_FMT.format('generator (test)',
Expand Down