1.简介
- 先上一张,我最喜欢的流程图
- G是一个生成式的网络,它接收一个随机的噪声z(随机数),通过这个噪声生成图像
- D是一个判别网络,判别一张图片是不是“真实的”。它的输入参数是x,x代表一张图片,输出D(x)代表x为真实图片的概率,如果为1,就代表100%是真实的图片,而输出为0,就代表不可能是真实的图片
2.接下来我们将以小例子的形式,了解GAN
2.1 定义我们的超参数
import torch
import torch.nn as nn
import numpy as np
import matplotlib.pyplot as plt
BATCH_SIZE = 64
LR_G = 0.0001
LR_D = 0.0001
N_IDEAS = 5
ART_CONPONENTS = 15
PAINT_POINTS = np.vstack([np.linspace(-1,1,ART_CONPONENTS) for _ in range(BATCH_SIZE)])
2.2 展示大师的画
plt.plot(PAINT_POINTS[0],2*np.power(PAINT_POINTS[0], 2) +1 , c='#74BCFF',lw=3,label='upper bound')
plt.plot(PAINT_POINTS[0],1*np.power(PAINT_POINTS[0], 2) +0 , c='#FF9359',lw=3,label='lower bound')
def artist_words():
a = np.random.uniform(1,2,BATCH_SIZE)[:, np.newaxis]
paintings = a * np.power(PAINT_POINTS,2)+(a-1)
plt.plot(np.linspace(-1,1,ART_CONPONENTS),paintings[1],label='really data',color='black',lw=3)
plt.legend(loc='upper right')
plt.show()
return paintings
2.3 定义对抗网络
G = nn.Sequential(
nn.Linear(N_IDEAS,128),
nn.ReLU(),
nn.Linear(128,ART_CONPONENTS),
)
D = nn.Sequential(
nn.Linear(ART_CONPONENTS,128),
nn.ReLU(),
nn.Linear(128,1),
nn.Sigmoid(),
)
opt_G = torch.optim.Adam(G.parameters(),lr=LR_G)
opt_D = torch.optim.Adam(D.parameters(),lr=LR_D)
2.4 训练GAN
for step in range(5000):
artist_paintings = artist_words()
G_ideas = torch.randn(BATCH_SIZE,N_IDEAS)
G_paintings = G(G_ideas)
prob_artist0 = D(artist_paintings)
prob_artist1 = D(G_paintings)
D_loss = - torch.mean(torch.log(prob_artist0) + torch.log(1-prob_artist1))
G_loss = torch.mean(torch.log(1- prob_artist1))
opt_D.zero_grad()
D_loss.backward(retain_graph=True)
opt_D.step()
opt_G.zero_grad()
G_loss.backward(retain_graph=True)
opt_G.step()
2.5 训练中的可视化
plt.ion()
for step in range(5000):
artist_paintings = artist_words()
G_ideas = torch.randn(BATCH_SIZE,N_IDEAS)
G_paintings = G(G_ideas)
prob_artist0 = D(artist_paintings)
prob_artist1 = D(G_paintings)
D_loss = - torch.mean(torch.log(prob_artist0) + torch.log(1-prob_artist1))
G_loss = torch.mean(torch.log(1- prob_artist1))
opt_D.zero_grad()
D_loss.backward(retain_graph=True)
opt_D.step()
opt_G.zero_grad()
G_loss.backward(retain_graph=True)
opt_G.step()
if step % 50 == 0:
plt.cla()
plt.plot(PAINT_POINTS[0],G_paintings.data.numpy()[0],c='#4AD631',lw=3,label='Generated painting')
plt.plot(PAINT_POINTS[0],2*np.power(PAINT_POINTS[0],2) + 1, c='#74BCFF', lw=3,label='upper bound')
plt.plot(PAINT_POINTS[0],1*np.power(PAINT_POINTS[0],2) + 0, c='#FF9359', lw=3,label='lower bound')
plt.text(-.5, 2.3, 'D accuracy=%.2f (0.5 for D to converge)'% prob_artist0.data.numpy().mean(),fontdict={'size':13})
plt.text(-.5, 2, 'D score=%.2f (-1.38 for G to converge)'% -D_loss.data.numpy(),fontdict={'size':13})
plt.ylim((0,3));plt.legend(loc='upper right', fontsize=10);plt.draw();plt.pause(0.02)
plt.ioff()
plt.show()