GAN的小测试

参考博客https://blog.csdn/Jason160918/article/details/78248066

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt

tf.set_random_seed(1)
np.random.seed(1)
# 设置超参数
batch_size = 64         # 批量大小
lr_g = 0.0001           # 生成器的学习率
lr_d = 0.0001           # 判别器的学习率
n_ideas = 5            # 认为这是生成艺术作品的几个想法（生成器）
art_components = 15     # 在画上画多少个点

#列表解析式代替了for循环，paint_points.shape=(64,15),
#np.vstack()默认逐行叠加（axis=0）
paint_points = np.vstack([np.linspace(-1, 1, art_components) for _ in range(batch_size)])

def artist_works():    
    
    #a为64个1到2均匀分布抽取的值，shape=(64,1)
    a = np.random.uniform(1, 2, size=batch_size)[:, np.newaxis]
    paintings = a * np.power(paint_points, 2) + (a-1)
    
    return paintings
with tf.variable_scope('Generator'):
    G_in = tf.placeholder(tf.float32, [None, n_ideas])          # 随机的ideals（来源于正态分布）
    G_l1 = tf.layers.dense(G_in, 256, tf.nn.relu)
    G_l2 = tf.layers.dense(G_l1, 128, tf.nn.relu)
    G_out = tf.layers.dense(G_l2, art_components)               # 得到15个点
with tf.variable_scope('Discriminator'):
    real_art = tf.placeholder(tf.float32, [None, art_components], name='real_in')   # 接受专家的画
    D_l0 = tf.layers.dense(real_art, 256, tf.nn.relu, name='l')
    D_l1 = tf.layers.dense(D_l0, 128, tf.nn.relu, name='l1')
    prob_artist0 = tf.layers.dense(D_l1, 1, tf.nn.sigmoid, name='out')              # 代入专家的画，判别器判断这副画来自于专家的概率
    # 再次利用生成器生成的15个点
    D_l2 = tf.layers.dense(G_out, 256, tf.nn.relu, name='l', reuse=True)  
    D_l3 = tf.layers.dense(D_l2, 128, tf.nn.relu, name='l1',reuse=True)          # 接受业余画家的画
    prob_artist1 = tf.layers.dense(D_l3, 1, tf.nn.sigmoid, name='out', reuse=True)  # 代入生成的画，判别器判断这副画来自于专家的概率
    
D_loss = -tf.reduce_mean(tf.log(prob_artist0) + tf.log(1-prob_artist1))#
G_loss = tf.reduce_mean(tf.log(1-prob_artist1))

train_D = tf.train.AdamOptimizer(lr_d).minimize(
    D_loss, var_list=tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, scope='Discriminator'))
train_G = tf.train.AdamOptimizer(lr_g).minimize(
    G_loss, var_list=tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, scope='Generator'))

sess = tf.Session()
sess.run(tf.global_variables_initializer())

plt.ion()   # 连续画图

for step in range(4000):
    artist_paintings = artist_works()           # 专家的画
    G_ideas = np.random.randn(batch_size, n_ideas)
    G_paintings, pa0, pa1, Dl = sess.run([G_out, prob_artist0, prob_artist1,D_loss, train_D, train_G], {G_in: G_ideas, real_art: artist_paintings})[:4]   # 训练和获取结果


    if step % 50 == 0:  # 每50步训练画一次图
        plt.cla()
        plt.plot(paint_points[0], G_paintings[0], c='#4AD631', lw=3, label='gen_paint',)
        plt.plot(paint_points[0], 2 * np.power(paint_points[0], 2) + 1, c='#74BCFF', lw=3, label='shangXian')
        plt.plot(paint_points[0], 1 * np.power(paint_points[0], 2) + 0, c='#FF9359', lw=3, label='xiaXian')
        plt.text(-.5, 2.3, 'D accuracy=%.2f (0.5 for D_real to converge)' % pa0.mean(), fontdict={'size': 15})
        plt.text(-.5, 2, 'D accuracy=%.2f (0.5 for D_fake to converge)' % pa1.mean(), fontdict={'size': 15})
        plt.text(-.5, 1.7, 'D score= %.2f (-1.38 for G to converge)' % -Dl, fontdict={'size': 15})
        plt.ylim((0, 3)); 
        plt.legend(loc='upper right', fontsize=12); 
        plt.draw(); 
        plt.pause(0.01)

plt.ioff()
plt.show()