文章目录
- 确定损失函数
- 生成器网络架构
不讲原理，从简单的代码一步步开始，学会怎么用、怎么设计损失函数即可。
确定损失函数
生成器的任务是生成足够以假乱真的数据，判别器的任务是分辨出哪些数据是真实的，哪些数据是假的。因此，对于判别器来讲，需要判别真伪，也就是true/false，从这个角度看，是个二分类问题。所以损失函数使用二类分类损失，即BCELoss。
import torch
import torch.nn as nn
adversarial_loss = nn.BCELoss()
生成器网络架构
这里使用纯线性网络作为生成器，得到的输出为[batch_size, np.pord(28*28)]
import torch.nn as nn
import numpy as np
class Generator(nn.Module):
def __init__(self):
super(Generator, self).__init__()
def block(in_features, out_features, normalization=True):
layers = [nn.Linear(in_features, out_features)]
if normalization:
layers.append(nn.BatchNorm1d(out_features, 0.8))
layers.append(nn.LeakyReLU(0.2))
return layers
self.model = nn.Sequential(
*block(100, 128, normalization=False),
*block(128, 256),
*block(256, 512),
*block(512, 1024),
nn.Linear(1024, int(np.prod((1, 28, 28)))), # generate a photo size, but in line mode.
nn.Tanh()
)
def forward(self, x):
return self.model(x)
判别器网络架构：
判别器的功能为判断出来哪一个是生成的图片，哪一个是真实的图片。对于生成的图片，我们希望判别器打上假的标签，对于真实的图片，我们希望判别器打上真的标签，因此，判别器的输出为一个数，即0或者1。
import torch.nn as nn
import numpy as np
class Disctiminator(nn.Module):
def __init__(self):
super(Disctiminator, self).__init__()
self.model = nn.Sequential(
nn.Linear(int(np.prod((1, 28, 28))), 512),
nn.LeakyReLU(0.2, inplace=True),
nn.Linear(512, 256),
nn.LeakyReLU(0.2, inplace=True),
nn.Linear(256, 1),
nn.Sigmoid()
)
def forward(self, x):
img_flat = x.view(x.size(0), -1)
validity = self.model(img_flat)
return validity
训练流程：
载入数据 — 训练 （生成图片 — 损失 — 反向传播） — 测试（这里没有加测试代码，可以照着训练代码改一下）
损失函数：生成器损失函数和判别器损失函数，两个损失函数分别进行反向传播，即生成器损失函数优化生成器，判别器损失函数优化判别器。
import torch
import torch.nn as nn
import argparse
import os
import numpy as np
from torch.utils.data import DataLoader, Dataset, dataset
from torchvision import datasets
from torchvision.transforms import transforms
from torchvision.utils import save_image
from torch.autograd import Variable
from models.generator import Generator
from models.dicsriminator import Disctiminator
os.makedirs('/home/sjr/gxj/study/data/mnist', exist_ok=True)
dataloader = DataLoader(datasets.MNIST('/home/sjr/gxj/study/data/mnist',
train=True, download=True,
transform=transforms.Compose([transforms.Resize(28), transforms.ToTensor(),
transforms.Normalize([0.5], [0.5])])
),
batch_size=64, shuffle=True, num_workers=4)
adversarial_loss = torch.nn.BCELoss()
generator = Generator()
discriminator = Disctiminator()
device = 'cuda:0' if torch.cuda.is_available() else 'cpu'
if torch.cuda.is_available():
adversarial_loss.cuda(device)
generator.cuda(device)
discriminator.cuda(device)
optimizer_G = torch.optim.Adam(generator.parameters(), lr=0.0002, betas=(0.5, 0.999))
optimizer_D = torch.optim.Adam(discriminator.parameters(), lr=0.0002, betas=(0.5, 0.999))
Tensor = torch.cuda.FloatTensor if torch.cuda.is_available() else torch.FloatTensor
for epoch in range(60):
for i, (imgs, _) in enumerate(dataloader):
valid = Tensor(imgs.size(0), 1).fill_(1.0)
fake = Tensor(imgs.size(0), 1).fill_(0.0)
real_imgs = Variable(imgs.type(Tensor))
optimizer_G.zero_grad()
z = Tensor(np.random.normal(0, 1, (imgs.shape[0], 100)))
gen_imgs = generator(z)
g_loss = adversarial_loss(discriminator(gen_imgs), valid)
g_loss.backward()
optimizer_G.step()
optimizer_D.zero_grad()
real_loss = adversarial_loss(discriminator(real_imgs), valid)
fake_loss = adversarial_loss(discriminator(gen_imgs.detach()), fake)
d_loss = (real_loss + fake_loss) / 2
d_loss.backward()
optimizer_D.step()
#
print(f"[Epoch {epoch}/{200}] [Batch {i}/{len(dataloader)}] [D loss: {d_loss.item()}] [G loss: {g_loss.item()}]")
#
if (epoch + 1) % 20 == 0:
save_image(gen_imgs.data[:25], f'images/{epoch+1}.png', nrow=5, normalize=True)
