使用 GAN 生成手写数字

以下是一个使用 GAN 生成手写数字的生成器的代码示例：

首先，我们需要导入必要的库：

import numpy as np
import matplotlib.pyplot as plt
from keras.models import Sequential
from keras.layers import Dense, Flatten, Reshape, Dropout
from keras.layers.convolutional import Conv2D, Conv2DTranspose
from keras.layers.normalization import BatchNormalization
from keras.layers.advanced_activations import LeakyReLU
from keras.optimizers import Adam
from keras.datasets import mnist

然后，我们需要加载 MNIST 数据集：

(x_train, _), (_, _) = mnist.load_data()
x_train = x_train / 127.5 - 1.
x_train = np.expand_dims(x_train, axis=3)

接下来，我们可以定义生成器和判别器的网络结构：

def build_generator():
    model = Sequential()
    model.add(Dense(256, input_dim=100))
    model.add(LeakyReLU(alpha=0.2))
    model.add(BatchNormalization())
    model.add(Dense(512))
    model.add(LeakyReLU(alpha=0.2))
    model.add(BatchNormalization())
    model.add(Dense(1024))
    model.add(LeakyReLU(alpha=0.2))
    model.add(BatchNormalization())
    model.add(Dense(28 * 28 * 1, activation='tanh'))
    model.add(Reshape((28, 28, 1)))
    return model

def build_discriminator():
    model = Sequential()
    model.add(Conv2D(64, (5, 5), strides=(2, 2), padding='same', input_shape=(28, 28, 1)))
    model.add(LeakyReLU(alpha=0.2))
    model.add(Dropout(0.3))
    model.add(Conv2D(128, (5, 5), strides=(2, 2), padding='same'))
    model.add(LeakyReLU(alpha=0.2))
    model.add(Dropout(0.3))
    model.add(Flatten())
    model.add(Dense(1, activation='sigmoid'))
    return model

然后，我们可以定义 GAN 的训练过程：

def build_gan(generator, discriminator):
    optimizer = Adam(lr=0.0002, beta_1=0.5)

    discriminator.trainable = False
    gan = Sequential()
    gan.add(generator)
    gan.add(discriminator)
    gan.compile(loss='binary_crossentropy', optimizer=optimizer)

    return gan

def train_gan(generator, discriminator, gan, epochs=100, batch_size=128):
    for epoch in range(epochs):
        for batch in range(x_train.shape[0] // batch_size):
            noise = np.random.normal(0, 1, size=(batch_size, 100))
            fake_images = generator.predict(noise)

            real_images = x_train[np.random.randint(0, x_train.shape[0], size=batch_size)]

            x = np.concatenate([real_images, fake_images])
            y = np.zeros(2 * batch_size)
            y[:batch_size] = 0.9

            discriminator.trainable = True
            discriminator.train_on_batch(x, y)

            noise = np.random.normal(0, 1, size=(batch_size, 100))
            y = np.ones(batch_size)
            discriminator.trainable = False
            gan.train_on_batch(noise, y)

        if (epoch + 1) % 10 == 0:
            print('Epoch {} completed'.format(epoch + 1))
            generate_and_save_images(generator, epoch + 1)

最后，我们可以定义一个函数来生成和保存图像：

def generate_and_save_images(generator, epoch):
    noise = np.random.normal(0, 1, size=(100, 100))
    generated_images = generator.predict(noise)
    generated_images = 0.5 * generated_images + 0.5
    fig, axs = plt.subplots(10, 10, figsize=(10, 10))
    cnt = 0
    for i in range(10):
        for j in range(10):
            axs[i, j].imshow(generated_images[cnt, :, :, 0], cmap='gray')
            axs[i, j].axis('off')
            cnt += 1
    fig.savefig('generated_images/{}.png'.format(epoch))
    plt.close()

现在我们可以训练 GAN 并生成手写数字图像：

generator = build_generator()
discriminator = build_discriminator()
gan = build_gan(generator, discriminator)
train_gan(generator, discriminator, gan, epochs=200)

训练完成后，生成的图像将保存在“generated_images”文件夹中。