LSGANs 简介与代码实战

1.介绍

  原始GAN（GAN 简介与代码实战_天竺街潜水的八角的博客-CSDN博客）生成的图片质量不高以及训练过程不稳定，LSGANs（最小二乘GAN）就是针对这两个缺陷进行改进，更加详细的介绍参考论文：Least Squares Generative Adversarial Networks

 

2.模型结构

 （a）是生成器，（b）是判别器，整个结构比较明朗，就不具体介绍了。

 

3.模型特点
    LSGANs是用来解决原始GAN的两个缺陷：生成的图片质量不高以及训练过程不稳定。

    1.以交叉熵作为损失，会使得生成器不会再优化那些被判别器识别为真实图片的生成图片，即使这些生成图片距离判别器的决策边界仍然很远，也就是距真实数据比较远。这意味着生成器的生成图片质量并不高。为什么生成器不再优化优化生成图片呢？是因为生成器已经完成我们为它设定的目标——尽可能地混淆判别器，所以交叉熵损失已经很小了（类似于逻辑斯回归分类）。而最小二乘就不一样了，要想最小二乘损失比较小，在混淆判别器的前提下还得让生成器把距离决策边界比较远的生成图片拉向决策边界（类似于曲线拟合），图中就很生动的反应了这个。

  2.以sigmoid交叉熵为损失函数，梯度容易为0，而最小二乘损失只是在x=1的时候，梯度为0

 

4.代码实现 keras

class LSGAN():
    def __init__(self):
        self.img_rows = 28
        self.img_cols = 28
        self.channels = 1
        self.img_shape = (self.img_rows, self.img_cols, self.channels)
        self.latent_dim = 100
 
        optimizer = Adam(0.0002, 0.5)
 
        # Build and compile the discriminator
        self.discriminator = self.build_discriminator()
        self.discriminator.compile(loss='mse',
            optimizer=optimizer,
            metrics=['accuracy'])
 
        # Build the generator
        self.generator = self.build_generator()
 
        # The generator takes noise as input and generated imgs
        z = Input(shape=(self.latent_dim,))
        img = self.generator(z)
 
        # For the combined model we will only train the generator
        self.discriminator.trainable = False
 
        # The valid takes generated images as input and determines validity
        valid = self.discriminator(img)
 
        # The combined model  (stacked generator and discriminator)
        # Trains generator to fool discriminator
        self.combined = Model(z, valid)
        # (!!!) Optimize w.r.t. MSE loss instead of crossentropy
        self.combined.compile(loss='mse', optimizer=optimizer)
 
    def build_generator(self):
 
        model = Sequential()
 
        model.add(Dense(256, input_dim=self.latent_dim))
        model.add(LeakyReLU(alpha=0.2))
        model.add(BatchNormalization(momentum=0.8))
        model.add(Dense(512))
        model.add(LeakyReLU(alpha=0.2))
        model.add(BatchNormalization(momentum=0.8))
        model.add(Dense(1024))
        model.add(LeakyReLU(alpha=0.2))
        model.add(BatchNormalization(momentum=0.8))
        model.add(Dense(np.prod(self.img_shape), activation='tanh'))
        model.add(Reshape(self.img_shape))
 
        model.summary()
 
        noise = Input(shape=(self.latent_dim,))
        img = model(noise)
 
        return Model(noise, img)
 
    def build_discriminator(self):
 
        model = Sequential()
 
        model.add(Flatten(input_shape=self.img_shape))
        model.add(Dense(512))
        model.add(LeakyReLU(alpha=0.2))
        model.add(Dense(256))
        model.add(LeakyReLU(alpha=0.2))
        # (!!!) No softmax
        model.add(Dense(1))
        model.summary()
 
        img = Input(shape=self.img_shape)
        validity = model(img)
 
        return Model(img, validity)
 
    def train(self, epochs, batch_size=128, sample_interval=50):
 
        # Load the dataset
        (X_train, _), (_, _) = mnist.load_data()
 
        # Rescale -1 to 1
        X_train = (X_train.astype(np.float32) - 127.5) / 127.5
        X_train = np.expand_dims(X_train, axis=3)
 
        # Adversarial ground truths
        valid = np.ones((batch_size, 1))
        fake = np.zeros((batch_size, 1))
 
        for epoch in range(epochs):
 
            # ---------------------
            #  Train Discriminator
            # ---------------------
 
            # Select a random batch of images
            idx = np.random.randint(0, X_train.shape[0], batch_size)
            imgs = X_train[idx]
 
            # Sample noise as generator input
            noise = np.random.normal(0, 1, (batch_size, self.latent_dim))
 
            # Generate a batch of new images
            gen_imgs = self.generator.predict(noise)
 
            # Train the discriminator
            d_loss_real = self.discriminator.train_on_batch(imgs, valid)
            d_loss_fake = self.discriminator.train_on_batch(gen_imgs, fake)
            d_loss = 0.5 * np.add(d_loss_real, d_loss_fake)
 
 
            # ---------------------
            #  Train Generator
            # ---------------------
 
            g_loss = self.combined.train_on_batch(noise, valid)
 
            # Plot the progress
            print ("%d [D loss: %f, acc.: %.2f%%] [G loss: %f]" % (epoch, d_loss[0], 100*d_loss[1], g_loss))
 
            # If at save interval => save generated image samples
            if epoch % sample_interval == 0:
                self.sample_images(epoch)
 
    def sample_images(self, epoch):
        r, c = 5, 5
        noise = np.random.normal(0, 1, (r * c, self.latent_dim))
        gen_imgs = self.generator.predict(noise)
 
        # Rescale images 0 - 1
        gen_imgs = 0.5 * gen_imgs + 0.5
 
        fig, axs = plt.subplots(r, c)
        cnt = 0
        for i in range(r):
            for j in range(c):
                axs[i,j].imshow(gen_imgs[cnt, :,:,0], cmap='gray')
                axs[i,j].axis('off')
                cnt += 1
        fig.savefig("images/mnist_%d.png" % epoch)
        plt.close()