对抗生成网络Gan变体集合 keras版本
一.ACGAN(Auxiliary Classifier GAN)
https://arxiv.org/abs/1610.09585
依旧有Generator,Discriminator,可使用MNSIT训练生成图片。
和DCGAN的不同:
1.增加了class类别标签参与训练,可以生成指定类别的图片
代码引用的《Web安全之强化学习与GAN》,位置:
https://github.com/duoergun0729/3book/tree/master/code/keras-acgan.py
生成器G代码:
def build_generator(latent_size):
cnn = Sequential()
cnn.add(Dense(1024, input_dim=latent_size, activation='relu'))
cnn.add(Dense(128 * 7 * 7, activation='relu'))
cnn.add(Reshape((128, 7, 7)))
cnn.add(UpSampling2D(size=(2, 2)))
cnn.add(Conv2D(256, (5, 5), padding="same", kernel_initializer="glorot_normal", activation="relu"))
cnn.add(UpSampling2D(size=(2, 2)))
cnn.add(Conv2D(128, (5, 5), padding="same", kernel_initializer="glorot_normal", activation="relu"))
cnn.add(Conv2D(1, (2, 2), padding="same", kernel_initializer="glorot_normal", activation="tanh"))
latent = Input(shape=(latent_size, ))
image_class = Input(shape=(1,), dtype='int32')
cls = Flatten()(Embedding(10, 100, embeddings_initializer="glorot_normal")(image_class))
#h = merge([latent, cls], mode='mul')
h=add([latent, cls])
fake_image = cnn(h)
return Model(inputs=[latent, image_class], outputs=[fake_image])
判别器D代码:
def build_discriminator():
cnn = Sequential()
cnn.add(Conv2D(32, (3, 3), padding="same", strides=(2, 2), input_shape=(1, 28, 28) ))
cnn.add(LeakyReLU())
cnn.add(Dropout(0.3))
cnn.add(Conv2D(64, (3, 3), padding="same", strides=(1, 1)))
cnn.add(LeakyReLU())
cnn.add(Dropout(0.3))
cnn.add(Conv2D(128, (3, 3), padding="same", strides=(2, 2)))
cnn.add(LeakyReLU())
cnn.add(Dropout(0.3))
cnn.add(Conv2D(256, (3, 3), padding="same", strides=(1, 1)))
cnn.add(LeakyReLU())
cnn.add(Dropout(0.3))
cnn.add(Flatten())
image = Input(shape=(1, 28, 28))
features = cnn(image)
fake = Dense(1, activation='sigmoid', name='generation')(features)
aux = Dense(10, activation='softmax', name='auxiliary')(features)
return Model(inputs=[image], outputs=[fake, aux])