原文链接:https://arxiv.org/pdf/1511.06434.pdf
简介
核心思想:本系列前面所有的GAN都是使用DNN结构来生成数据的,但是本文开始引入了在图像处理中的卷积结构。
如上图所示,本文中GAN的生成器中是没有用FC(全连接层)的,而是使用反卷积来逐步扩大H和W两个维度,并逐步缩小C维度,生成最终结果图。其中,第一步将一百维的随机噪声投影并变形为4x4x1024的特征图,具体的做法还是使用了FC层将输出设为8192的向量转换为4x4x1024的特征图。其次,本文中还提到了将GAN用于无监督学习方法中,并且对生成器插入噪声与生成结果的关系进行了探究,还对判别器内部特征进行了可视化。
基础结构
上采样
在应用在计算机视觉的深度学习领域,由于输入图像通过卷积神经网络(CNN)提取特征后,输出的尺寸往往会变小,而有时我们需要将图像恢复到原来的尺寸以便进行进一步的计算,这个采用扩大图像尺寸,实现图像由小分辨率到大分辨率的映射的操作,叫做上采样(Upsample)。
反卷积
反卷积,也叫转置卷积,它并不是正向卷积的完全逆过程,用一句话来解释:
反卷积是一种特殊的正向卷积,先按照一定的比例通过补0来扩大输入图像的尺寸,接着旋转卷积核,再进行正向卷积。
无监督学习
此文章根据GAN网络不需要标签的性质,构想出了一种无监督学习策略:当GAN网络的生成器可以生成较为真实的数据时,同时也证明其判别器也学习到了数据相关的表示,那么我们将判别器最后的FC层去掉即可获得一个数据的特征提取器。为了验证此特征提取器是否能有效的提取到数据的特征,我们可以用提取出的特征进行分类,然后对比实际的分类标签来评估特征提取器的性能。文中作者分别使用了CIFAR-10与SVHN数据集进行了实验,结果分别如下:
可视化
生成器隐状态研究
作者通过改变输入噪声z中的某几位数值,以控制变量法来观察最终生成图片的形态。发现其中特定的数位可以控制生成图片中的特定形态表示。根据数位数值大小的变化,其生成的图片结果如下:
甚至于对于输入的向量可以通过“运算”来控制图片的“运算”:
判别器可视化
如上图所示,左侧的特征图是没有训练时判别器的内部卷积可视化,而右边的时经过训练之后判别器的内部卷积可视化。可以明显的看到,判别器的确学到了数据表现出的特征,也印证了GAN网络可以运用到无监督学习中。
代码与实践
参考链接(https://github.com/eriklindernoren/PyTorch-GAN/blob/master/implementations/dcgan/dcgan.py)
import argparse
import os
import numpy as np
import math
import torchvision.transforms as transforms
from torchvision.utils import save_image
from torch.utils.data import DataLoader
from torchvision import datasets
from torch.autograd import Variable
import torch.nn as nn
import torch.nn.functional as F
import torch
os.makedirs("images", exist_ok=True)
parser = argparse.ArgumentParser()
parser.add_argument("--n_epochs", type=int, default=200, help="number of epochs of training")
parser.add_argument("--batch_size", type=int, default=64, help="size of the batches")
parser.add_argument("--lr", type=float, default=0.0002, help="adam: learning rate")
parser.add_argument("--b1", type=float, default=0.5, help="adam: decay of first order momentum of gradient")
parser.add_argument("--b2", type=float, default=0.999, help="adam: decay of first order momentum of gradient")
parser.add_argument("--n_cpu", type=int, default=8, help="number of cpu threads to use during batch generation")
parser.add_argument("--latent_dim", type=int, default=100, help="dimensionality of the latent space")
parser.add_argument("--img_size", type=int, default=32, help="size of each image dimension")
parser.add_argument("--channels", type=int, default=1, help="number of image channels")
parser.add_argument("--sample_interval", type=int, default=400, help="interval between image sampling")
opt = parser.parse_args()
print(opt)
cuda = True if torch.cuda.is_available() else False
def weights_init_normal(m):
classname = m.__class__.__name__
if classname.find("Conv") != -1:
torch.nn.init.normal_(m.weight.data, 0.0, 0.02)
elif classname.find("BatchNorm2d") != -1:
torch.nn.init.normal_(m.weight.data, 1.0, 0.02)
torch.nn.init.constant_(m.bias.data, 0.0)
class Generator(nn.Module):
def __init__(self):
super(Generator, self).__init__()
self.init_size = opt.img_size // 4
self.l1 = nn.Sequential(nn.Linear(opt.latent_dim, 128 * self.init_size ** 2))
self.conv_blocks = nn.Sequential(
nn.BatchNorm2d(128),
nn.Upsample(scale_factor=2),
nn.Conv2d(128, 128, 3, stride=1, padding=1),
nn.BatchNorm2d(128, 0.8),
nn.LeakyReLU(0.2, inplace=True),
nn.Upsample(scale_factor=2),
nn.Conv2d(128, 64, 3, stride=1, padding=1),
nn.BatchNorm2d(64, 0.8),
nn.LeakyReLU(0.2, inplace=True),
nn.Conv2d(64, opt.channels, 3, stride=1, padding=1),
nn.Tanh(),
)
def forward(self, z):
out = self.l1(z)
out = out.view(out.shape[0], 128, self.init_size, self.init_size)
img = self.conv_blocks(out)
return img
class Discriminator(nn.Module):
def __init__(self):
super(Discriminator, self).__init__()
def discriminator_block(in_filters, out_filters, bn=True):
block = [nn.Conv2d(in_filters, out_filters, 3, 2, 1), nn.LeakyReLU(0.2, inplace=True), nn.Dropout2d(0.25)]
if bn:
block.append(nn.BatchNorm2d(out_filters, 0.8))
return block
self.model = nn.Sequential(
*discriminator_block(opt.channels, 16, bn=False),
*discriminator_block(16, 32),
*discriminator_block(32, 64),
*discriminator_block(64, 128),
)
# The height and width of downsampled image
ds_size = opt.img_size // 2 ** 4
self.adv_layer = nn.Sequential(nn.Linear(128 * ds_size ** 2, 1), nn.Sigmoid())
def forward(self, img):
out = self.model(img)
out = out.view(out.shape[0], -1)
validity = self.adv_layer(out)
return validity
# Loss function
adversarial_loss = torch.nn.BCELoss()
# Initialize generator and discriminator
generator = Generator()
discriminator = Discriminator()
if cuda:
generator.cuda()
discriminator.cuda()
adversarial_loss.cuda()
# Initialize weights
generator.apply(weights_init_normal)
discriminator.apply(weights_init_normal)
# Configure data loader
os.makedirs("../../data/mnist", exist_ok=True)
dataloader = torch.utils.data.DataLoader(
datasets.MNIST(
"../../data/mnist",
train=True,
download=True,
transform=transforms.Compose(
[transforms.Resize(opt.img_size), transforms.ToTensor(), transforms.Normalize([0.5], [0.5])]
),
),
batch_size=opt.batch_size,
shuffle=True,
)
# Optimizers
optimizer_G = torch.optim.Adam(generator.parameters(), lr=opt.lr, betas=(opt.b1, opt.b2))
optimizer_D = torch.optim.Adam(discriminator.parameters(), lr=opt.lr, betas=(opt.b1, opt.b2))
Tensor = torch.cuda.FloatTensor if cuda else torch.FloatTensor
# ----------
# Training
# ----------
for epoch in range(opt.n_epochs):
for i, (imgs, _) in enumerate(dataloader):
# Adversarial ground truths
valid = Variable(Tensor(imgs.shape[0], 1).fill_(1.0), requires_grad=False)
fake = Variable(Tensor(imgs.shape[0], 1).fill_(0.0), requires_grad=False)
# Configure input
real_imgs = Variable(imgs.type(Tensor))
# -----------------
# Train Generator
# -----------------
optimizer_G.zero_grad()
# Sample noise as generator input
z = Variable(Tensor(np.random.normal(0, 1, (imgs.shape[0], opt.latent_dim))))
# Generate a batch of images
gen_imgs = generator(z)
# Loss measures generator's ability to fool the discriminator
g_loss = adversarial_loss(discriminator(gen_imgs), valid)
g_loss.backward()
optimizer_G.step()
# ---------------------
# Train Discriminator
# ---------------------
optimizer_D.zero_grad()
# Measure discriminator's ability to classify real from generated samples
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(
"[Epoch %d/%d] [Batch %d/%d] [D loss: %f] [G loss: %f]"
% (epoch, opt.n_epochs, i, len(dataloader), d_loss.item(), g_loss.item())
)
batches_done = epoch * len(dataloader) + i
if batches_done % opt.sample_interval == 0:
save_image(gen_imgs.data[:25], "images/%d.png" % batches_done, nrow=5, normalize=True)
minist测试
可见使用了卷积结构后,生成器生成的图片收敛速度更快,且图片内容的结构也更加清晰,并且是在使用原始GAN的LOSS函数下,如果引入“w散度”的话,强强联合是否能生成更加好的结果。