模型简介
GAN模型的核心在于提出了通过对抗过程来估计生成模型这一全新框架。在这个框架中，将会同时训练两个模型——捕捉数据分布的生成模型G和估计样本是否来自训练数据的判别模型D 。
在训练过程中，生成器会不断尝试通过生成更好的假图像来骗过判别器，而判别器在这过程中也会逐步提升判别能力。这种博弈的平衡点是，当生成器生成的假图像和训练数据图像的分布完全一致时，判别器拥有50%的真假判断置信度。
数据集
使用MNIST手写数字数据集
数据集下载和加载
# 数据下载
from download import downloadurl = "https://mindspore-website.obs.cn-north-4.myhuaweicloud.com/notebook/datasets/MNIST_Data.zip"
download(url, ".", kind="zip", replace=True)import numpy as np
import mindspore.dataset as dsbatch_size = 64
latent_size = 100 # 隐码的长度train_dataset = ds.MnistDataset(dataset_dir='./MNIST_Data/train')
test_dataset = ds.MnistDataset(dataset_dir='./MNIST_Data/test')def data_load(dataset):dataset1 = ds.GeneratorDataset(dataset, ["image", "label"], shuffle=True, python_multiprocessing=False,num_samples=10000)# 数据增强mnist_ds = dataset1.map(operations=lambda x: (x.astype("float32"), np.random.normal(size=latent_size).astype("float32")),output_columns=["image", "latent_code"])mnist_ds = mnist_ds.project(["image", "latent_code"])# 批量操作mnist_ds = mnist_ds.batch(batch_size, True)return mnist_dsmnist_ds = data_load(train_dataset)iter_size = mnist_ds.get_dataset_size()
print('Iter size: %d' % iter_size)隐码构造
为了跟踪生成器的学习进度，我们在训练的过程中的每轮迭代结束后，将一组固定的遵循高斯分布的隐码输入到生成器中，通过固定隐码所生成的图像效果来评估生成器的好坏。
import random
import numpy as np
from mindspore import Tensor
from mindspore.common import dtype# 利用随机种子创建一批隐码
np.random.seed(2323)
test_noise = Tensor(np.random.normal(size=(25, 100)), dtype.float32)
random.shuffle(test_noise)模型构建
生成器
from mindspore import nn
import mindspore.ops as opsimg_size = 28 # 训练图像长（宽）class Generator(nn.Cell):def __init__(self, latent_size, auto_prefix=True):super(Generator, self).__init__(auto_prefix=auto_prefix)self.model = nn.SequentialCell()# [N, 100] -> [N, 128]# 输入一个100维的0～1之间的高斯分布，然后通过第一层线性变换将其映射到256维self.model.append(nn.Dense(latent_size, 128))self.model.append(nn.ReLU())# [N, 128] -> [N, 256]self.model.append(nn.Dense(128, 256))self.model.append(nn.BatchNorm1d(256))self.model.append(nn.ReLU())# [N, 256] -> [N, 512]self.model.append(nn.Dense(256, 512))self.model.append(nn.BatchNorm1d(512))self.model.append(nn.ReLU())# [N, 512] -> [N, 1024]self.model.append(nn.Dense(512, 1024))self.model.append(nn.BatchNorm1d(1024))self.model.append(nn.ReLU())# [N, 1024] -> [N, 784]# 经过线性变换将其变成784维self.model.append(nn.Dense(1024, img_size * img_size))# 经过Tanh激活函数是希望生成的假的图片数据分布能够在-1～1之间self.model.append(nn.Tanh())def construct(self, x):img = self.model(x)return ops.reshape(img, (-1, 1, 28, 28))net_g = Generator(latent_size)
net_g.update_parameters_name('generator')判别器
# 判别器
class Discriminator(nn.Cell):def __init__(self, auto_prefix=True):super().__init__(auto_prefix=auto_prefix)self.model = nn.SequentialCell()# [N, 784] -> [N, 512]self.model.append(nn.Dense(img_size * img_size, 512)) # 输入特征数为784，输出为512self.model.append(nn.LeakyReLU()) # 默认斜率为0.2的非线性映射激活函数# [N, 512] -> [N, 256]self.model.append(nn.Dense(512, 256)) # 进行一个线性映射self.model.append(nn.LeakyReLU())# [N, 256] -> [N, 1]self.model.append(nn.Dense(256, 1))self.model.append(nn.Sigmoid()) # 二分类激活函数，将实数映射到[0,1]def construct(self, x):x_flat = ops.reshape(x, (-1, img_size * img_size))return self.model(x_flat)net_d = Discriminator()
net_d.update_parameters_name('discriminator')损失函数和优化器
lr = 0.0002 # 学习率# 损失函数
adversarial_loss = nn.BCELoss(reduction='mean')# 优化器
optimizer_d = nn.Adam(net_d.trainable_params(), learning_rate=lr, beta1=0.5, beta2=0.999)
optimizer_g = nn.Adam(net_g.trainable_params(), learning_rate=lr, beta1=0.5, beta2=0.999)
optimizer_g.update_parameters_name('optim_g')
optimizer_d.update_parameters_name('optim_d')模型训练
第一部分是训练判别器。训练判别器的目的是最大程度地提高判别图像真伪的概率。
第二部分是训练生成器。以产生更好的虚拟图像。
import os
import time
import matplotlib.pyplot as plt
import mindspore as ms
from mindspore import Tensor, save_checkpointtotal_epoch = 12 # 训练周期数
batch_size = 64 # 用于训练的训练集批量大小# 加载预训练模型的参数
pred_trained = False
pred_trained_g = './result/checkpoints/Generator99.ckpt'
pred_trained_d = './result/checkpoints/Discriminator99.ckpt'checkpoints_path = "./result/checkpoints" # 结果保存路径
image_path = "./result/images" # 测试结果保存路径# 生成器计算损失过程
def generator_forward(test_noises):fake_data = net_g(test_noises)fake_out = net_d(fake_data)loss_g = adversarial_loss(fake_out, ops.ones_like(fake_out))return loss_g# 判别器计算损失过程
def discriminator_forward(real_data, test_noises):fake_data = net_g(test_noises)fake_out = net_d(fake_data)real_out = net_d(real_data)real_loss = adversarial_loss(real_out, ops.ones_like(real_out))fake_loss = adversarial_loss(fake_out, ops.zeros_like(fake_out))loss_d = real_loss + fake_lossreturn loss_d# 梯度方法
grad_g = ms.value_and_grad(generator_forward, None, net_g.trainable_params())
grad_d = ms.value_and_grad(discriminator_forward, None, net_d.trainable_params())def train_step(real_data, latent_code):# 计算判别器损失和梯度loss_d, grads_d = grad_d(real_data, latent_code)optimizer_d(grads_d)loss_g, grads_g = grad_g(latent_code)optimizer_g(grads_g)return loss_d, loss_g# 保存生成的test图像
def save_imgs(gen_imgs1, idx):for i3 in range(gen_imgs1.shape[0]):plt.subplot(5, 5, i3 + 1)plt.imshow(gen_imgs1[i3, 0, :, :] / 2 + 0.5, cmap="gray")plt.axis("off")plt.savefig(image_path + "/test_{}.png".format(idx))# 设置参数保存路径
os.makedirs(checkpoints_path, exist_ok=True)
# 设置中间过程生成图片保存路径
os.makedirs(image_path, exist_ok=True)net_g.set_train()
net_d.set_train()# 储存生成器和判别器loss
losses_g, losses_d = [], []for epoch in range(total_epoch):start = time.time()for (iter, data) in enumerate(mnist_ds):start1 = time.time()image, latent_code = dataimage = (image - 127.5) / 127.5 # [0, 255] -> [-1, 1]image = image.reshape(image.shape[0], 1, image.shape[1], image.shape[2])d_loss, g_loss = train_step(image, latent_code)end1 = time.time()if iter % 10 == 10:print(f"Epoch:[{int(epoch):>3d}/{int(total_epoch):>3d}], "f"step:[{int(iter):>4d}/{int(iter_size):>4d}], "f"loss_d:{d_loss.asnumpy():>4f} , "f"loss_g:{g_loss.asnumpy():>4f} , "f"time:{(end1 - start1):>3f}s, "f"lr:{lr:>6f}")end = time.time()print("time of epoch {} is {:.2f}s".format(epoch + 1, end - start))losses_d.append(d_loss.asnumpy())losses_g.append(g_loss.asnumpy())# 每个epoch结束后，使用生成器生成一组图片gen_imgs = net_g(test_noise)save_imgs(gen_imgs.asnumpy(), epoch)# 根据epoch保存模型权重文件if epoch % 1 == 0:save_checkpoint(net_g, checkpoints_path + "/Generator%d.ckpt" % (epoch))save_checkpoint(net_d, checkpoints_path + "/Discriminator%d.ckpt" % (epoch))
生成图像：
效果展示
描绘D和G损失与训练迭代的关系图：
plt.figure(figsize=(6, 4))
plt.title("Generator and Discriminator Loss During Training")
plt.plot(losses_g, label="G", color='blue')
plt.plot(losses_d, label="D", color='orange')
plt.xlim(-5,15)
plt.ylim(0, 3.5)
plt.xlabel("iterations")
plt.ylabel("Loss")
plt.legend()
plt.show()
可视化训练过程中通过隐向量生成的图像。
import cv2
import matplotlib.animation as animation# 将训练过程中生成的测试图转为动态图
image_list = []
for i in range(total_epoch):image_list.append(cv2.imread(image_path + "/test_{}.png".format(i), cv2.IMREAD_GRAYSCALE))
show_list = []
fig = plt.figure(dpi=70)
for epoch in range(0, len(image_list), 5):plt.axis("off")show_list.append([plt.imshow(image_list[epoch], cmap='gray')])ani = animation.ArtistAnimation(fig, show_list, interval=1000, repeat_delay=1000, blit=True)
ani.save('train_test.gif', writer='pillow', fps=1)
可见随着训练次数的增多，图像质量也越来越好。
模型推理
通过加载生成器网络模型参数文件来生成图像，代码如下：
import mindspore as ms# test_ckpt = './result/checkpoints/Generator199.ckpt'# parameter = ms.load_checkpoint(test_ckpt)
# ms.load_param_into_net(net_g, parameter)
# 模型生成结果
test_data = Tensor(np.random.normal(0, 1, (25, 100)).astype(np.float32))
images = net_g(test_data).transpose(0, 2, 3, 1).asnumpy()
# 结果展示
fig = plt.figure(figsize=(3, 3), dpi=120)
for i in range(25):fig.add_subplot(5, 5, i + 1)plt.axis("off")plt.imshow(images[i].squeeze(), cmap="gray")
plt.show()
总结
生成对抗网络GAN通过生成器和判别器之间的对抗训练，从而获得高质量的数据扩充。
