gan python实现

GAN是生成式对抗网络（Generative Adversarial Networks）的缩写，是一种用于生成模拟数据的机器学习模型。下面是一个简单的GAN Python实现示例:

import numpy as np
import matplotlib.pyplot as plt
from keras.layers import Dense, Input
from keras.models import Model
from keras.optimizers import Adam

# 生成器模型
def build_generator(latent_dim):
    inputs = Input(shape=(latent_dim,))
    x = Dense(128, activation='relu')(inputs)
    x = Dense(256, activation='relu')(x)
    x = Dense(512, activation='relu')(x)
    outputs = Dense(784, activation='sigmoid')(x)
    model = Model(inputs=inputs, outputs=outputs)
    return model

# 判别器模型
def build_discriminator():
    inputs = Input(shape=(784,))
    x = Dense(512, activation='relu')(inputs)
    x = Dense(256, activation='relu')(x)
    x = Dense(128, activation='relu')(x)
    outputs = Dense(1, activation='sigmoid')(x)
    model = Model(inputs=inputs, outputs=outputs)
    return model

# GAN模型
def build_gan(generator, discriminator):
    discriminator.trainable = False
    inputs = Input(shape=(100,))
    gan_outputs = discriminator(generator(inputs))
    model = Model(inputs=inputs, outputs=gan_outputs)
    model.compile(loss='binary_crossentropy', optimizer=Adam(lr=0.0002, beta_1=0.5))
    return model

# 加载数据集
(X_train, _), (_, _) = mnist.load_data()
X_train = X_train.reshape(X_train.shape[0], 784)
X_train = X_train.astype('float32') / 255

# 构建模型
generator = build_generator(100)
discriminator = build_discriminator()
gan = build_gan(generator, discriminator)

# 训练GAN模型
for epoch in range(1000):
    noise = np.random.normal(0, 1, (100,)) # 生成噪声
    fake_images = generator.predict(noise) # 生成假图片
    real_images = X_train[np.random.randint(0, X_train.shape[0], size=100)] # 从真实数据集中随机取出100张图片

    # 训练判别器
    discriminator.trainable = True
    discriminator.train_on_batch(real_images, np.ones((100,)))
    discriminator.train_on_batch(fake_images, np.zeros((100,)))

    # 训练生成器
    noise = np.random.normal(0, 1, (100,))
    discriminator.trainable = False
    gan.train_on_batch(noise, np.ones((100,)))

    # 每100轮输出一次生成的图片
    if epoch % 100 == 0:
        print('Epoch %d' % epoch)
        plt.imshow(fake_images[0].reshape(28, 28), cmap='gray')
        plt.show()

以上是一个基本的GAN Python实现示例，它使用了Keras深度学习框架和MNIST手写数字数据集来生成手写数字图像。该示例中的生成器和判别器模型都是基于全连接层的简单模型，可以根据需求进行更改和优化。