（大结局）左右互搏:生成型对抗性网络的强大威力

import  keras
from  keras  import  layers
import numpy as np

#输入generator网络的随机向量长度
latent_dim = 32
#generator输出格式为[32, 32 , 3]的数组，它对应一张图片
height = 32
width = 32
channels = 3

generator_input = keras.Input(shape = (latent_dim, ))

x = layers.Dense(128 * 16 * 16)(generator_input)
x = layers.LeakyReLU()(x)
x = layers.Reshape((16, 16, 128))(x)
x = layers.Conv2D(256, 5, padding= 'same')(x)
#我们使用激活函数LeakyReLu而不是以前的Relu，前者有利于网络训练时趋于稳定
x = layers.LeakyReLU()(x)

#卷积网络层
x = layers.Conv2DTranspose(256, 4, strides = 2, padding = 'same')(x)
x = layers.LeakyReLU()(x)

x = layers.Conv2D(256, 5, padding = 'same')(x)
x = layers.LeakyReLU()(x)

x = layers.Conv2D(256, 5, padding = 'same')(x)
x = layers.LeakyReLU()(x)

#使用激活函数tanh而不是sigmoid，因为它有利于网络在训练时趋于稳定
x = layers.Conv2D(channels, 7, activation = 'tanh', padding = 'same')(x)

generator = keras.models.Model(generator_input, x)
generator.summary()

#generator的输出就是discriminator的输入
discriminator_input = layers.Input(shape=(height, width, channels))
x = layers.Conv2D(128, 3)(discriminator_input)
x = layers.LeakyReLU()(x)
x = layers.Conv2D(128, 4, strides = 2)(x)
x = layers.LeakyReLU()(x)
x = layers.Conv2D(128, 4, strides = 2)(x)
x = layers.LeakyReLU()(x)
x = layers.Conv2D(128, 4, strides = 2)(x)
x = layers.LeakyReLU()(x)
x = layers.Flatten()(x)
#增加Dropout有利于网络训练时趋于稳定
x = layers.Dropout(0.4)(x)
x = layers.Dense(1, activation = 'sigmoid')(x)

discriminator = keras.models.Model(discriminator_input, x)
discriminator.summary()

discriminator_optimizer = keras.optimizers.RMSprop(lr = 0.0008, 
                                                   #因为网络训练时状态变化很剧烈，因此我们限定参数变化范围减少状态摇摆
                                                   clipvalue = 1.0,
                                                   #学习率也要不断变化以适应网络状态的改变
                                                   decay = 1e-8
                                                  )
#discriminator判断generator构造的图片是否为真
discriminator.compile(optimizer = discriminator_optimizer, 
                     loss = 'binary_crossentropy')

'''
我们把generator和discriminator连成一个整体，在对整体进行训练时，
只更改generator网络的参数，discriminator的参数保持不变
'''
discriminator.trainable = False
gan_input = keras.Input(shape = (latent_dim, ))

#将两个网络衔接在一起
gan_output = discriminator(generator(gan_input))

gan = keras.models.Model(gan_input, gan_output)

gan_optimizer = keras.optimizers.RMSprop(lr = 0.0004, clipvalue = 1.0,
                                        decay = 1e-8)
gan.compile(optimizer = gan_optimizer, loss = 'binary_crossentropy')

import os
from keras.preprocessing import image

(x_train, y_train), (_, _) = keras.datasets.cifar10.load_data()
#选出所有青蛙图片
x_train = x_train[y_train.flatten() == 6]

x_train = x_train.reshape((x_train.shape[0], ) + 
                         (height, width, channels)).astype('float32') / 255.
iterations = 10000
batch_size = 20
save_dir = '/content/gdrive/My Drive/gen_imgs'

start = 0
for step in range(iterations):
  random_latent_vectors = np.random.normal(size = (batch_size, latent_dim))
  #让generator产生对应图片的二维数组
  generated_images = generator.predict(random_latent_vectors)

  stop = start + batch_size
  real_images = x_train[start : stop]
  combined_images = np.concatenate([generated_images, real_images])

  labels = np.concatenate([np.ones((batch_size, 1)), np.zeros((batch_size, 1))])

  #这是一个让网络训练趋于稳定的小技巧，就是将给标签添加随机化噪音
  labels += 0.05 * np.random.random(labels.shape)
  #先训练discriminator识别真假图片
  d_loss = discriminator.train_on_batch(combined_images, labels)

  random_latent_vectors = np.random.normal(size = (batch_size, latent_dim))
  misleading_targets = np.zeros((batch_size, 1))
  #根据discriminator的反馈让generator改进自身参数
  a_loss = gan.train_on_batch(random_latent_vectors, misleading_targets)

  start += batch_size
  if start > len(x_train) - batch_size:
    start = 0

  if step % 100 == 0:
    gan.save_weights('gan.h5')
    print('discriminator loss: ', d_loss)
    print('adversarial loss: ', a_loss)
    img = image.array_to_img(generated_images[0] * 255. , scale = False)
    img.save(os.path.join(save_dir, 'generated_frog' + str(step) + '.png'))

    img = image.array_to_img(real_images[0] * 255., scale = False)
    img.save(os.path.join(save_dir, 'real_frog' + str(step) + '.png'))