Pytorch | 利用NI-FGSM针对CIFAR10上的ResNet分类器进行对抗攻击
- CIFAR数据集
- NI-FGSM介绍
- 背景
- 算法流程
- NI-FGSM代码实现
- NI-FGSM算法实现
- 攻击效果
- 代码汇总
- nifgsm.py
- train.py
- advtest.py
之前已经针对CIFAR10训练了多种分类器:
Pytorch | 从零构建AlexNet对CIFAR10进行分类
Pytorch | 从零构建Vgg对CIFAR10进行分类
Pytorch | 从零构建GoogleNet对CIFAR10进行分类
Pytorch | 从零构建ResNet对CIFAR10进行分类
Pytorch | 从零构建MobileNet对CIFAR10进行分类
Pytorch | 从零构建EfficientNet对CIFAR10进行分类
Pytorch | 从零构建ParNet对CIFAR10进行分类
本篇文章我们使用Pytorch实现NI-FGSM对CIFAR10上的ResNet分类器进行攻击.
CIFAR数据集
CIFAR-10数据集是由加拿大高级研究所(CIFAR)收集整理的用于图像识别研究的常用数据集,基本信息如下:
- 数据规模:该数据集包含60,000张彩色图像,分为10个不同的类别,每个类别有6,000张图像。通常将其中50,000张作为训练集,用于模型的训练;10,000张作为测试集,用于评估模型的性能。
- 图像尺寸:所有图像的尺寸均为32×32像素,这相对较小的尺寸使得模型在处理该数据集时能够相对快速地进行训练和推理,但也增加了图像分类的难度。
- 类别内容:涵盖了飞机(plane)、汽车(car)、鸟(bird)、猫(cat)、鹿(deer)、狗(dog)、青蛙(frog)、马(horse)、船(ship)、卡车(truck)这10个不同的类别,这些类别都是现实世界中常见的物体,具有一定的代表性。
下面是一些示例样本:
NI-FGSM介绍
NI-FGSM(Nesterov Iterative Fast Gradient Sign Method)即涅斯捷罗夫迭代快速梯度符号法,是一种在对抗攻击领域中对FGSM进行改进的迭代攻击算法,以下是其详细介绍:
背景
- 传统的FGSM及其一些迭代改进版本如I-FGSM等,在生成对抗样本时存在一些局限性,例如可能会在迭代过程中陷入局部最优,导致攻击效果不够理想或生成的对抗样本转移性较差。NI-FGSM借鉴了优化算法中的Nesterov加速梯度法的思想,旨在更有效地利用梯度信息,提高攻击的效率和效果。
算法流程
- 初始化
- 设定参数:确定最大扰动幅度 ϵ \epsilon ϵ、迭代次数 T T T、步长 α = ϵ / T \alpha = \epsilon / T α=ϵ/T、衰减因子 μ \mu μ。
- 初始化变量:令初始累积梯度 g 0 = 0 g_0 = 0 g0=0,初始对抗样本 x 0 a d v = x x_0^{adv}=x x0adv=x( x x x 为原始图像)。
- 迭代过程
- 对于每次迭代
t
=
0
t = 0
t=0到
T
−
1
T - 1
T−1:
- 计算跳跃点 x t n e s = x t a d v + α ⋅ μ ⋅ g t x_t^{nes}=x_t^{adv}+\alpha \cdot \mu \cdot g_t xtnes=xtadv+α⋅μ⋅gt。
- 计算跳跃点 x t n e s x_t^{nes} xtnes 在模型中的梯度 ∇ x J ( x t n e s , y t r u e ) \nabla_{x} J\left(x_t^{nes }, y^{true }\right) ∇xJ(xtnes,ytrue)。
- 累积梯度更新: g t + 1 = μ ⋅ g t + ∇ x J ( x t n e s , y t r u e ) ∥ ∇ x J ( x t n e s , y t r u e ) ∥ 1 g_{t + 1}=\mu \cdot g_t+\frac{\nabla_{x} J\left(x_t^{nes }, y^{true }\right)}{\left\| \nabla_{x} J\left(x_t^{nes}, y^{true }\right)\right\| _{1}} gt+1=μ⋅gt+∥∇xJ(xtnes,ytrue)∥1∇xJ(xtnes,ytrue)。
- 更新对抗样本: x t + 1 a d v = C l i p x ϵ { x t a d v + α ⋅ s i g n ( g t + 1 ) } x_{t + 1}^{adv}=Clip_{x}^{\epsilon}\left\{x_t^{adv}+\alpha \cdot sign\left(g_{t + 1}\right)\right\} xt+1adv=Clipxϵ{xtadv+α⋅sign(gt+1)},其中 C l i p x ϵ Clip_{x}^{\epsilon} Clipxϵ 函数将生成的对抗样本限制在原始图像 x x x 的 ϵ \epsilon ϵ 邻域内。
- 对于每次迭代
t
=
0
t = 0
t=0到
T
−
1
T - 1
T−1:
- 输出结果:返回最终生成的对抗样本 x T a d v x_T^{adv} xTadv。
在整个算法流程中,通过引入Nesterov加速梯度的思想,在每次迭代计算梯度之前先进行跳跃操作,从而利用先前累积梯度的信息来更有效地更新对抗样本,使得生成的对抗样本具有更好的转移性,能够在不同模型上保持较高的攻击成功率。
NI-FGSM代码实现
NI-FGSM算法实现
import torch
import torch.nn as nn
def NI_FGSM(model, criterion, original_images, labels, epsilon, num_iterations=10, decay=1):
"""
NI-FGSM (Nesterov Iterative Fast Gradient Sign Method)
参数:
- model: 要攻击的模型
- criterion: 损失函数
- original_images: 原始图像
- labels: 原始图像的标签
- epsilon: 最大扰动幅度
- num_iterations: 迭代次数
- decay: 动量衰减因子
"""
# alpha 每次迭代步长
alpha = epsilon / num_iterations
# 复制原始图像作为初始的对抗样本
perturbed_images = original_images.clone().detach().requires_grad_(True)
momentum = torch.zeros_like(original_images).detach().to(original_images.device)
for _ in range(num_iterations):
# 先沿先前累积梯度的方向进行跳跃
nes_images = perturbed_images + alpha * decay * momentum
nes_images = nes_images.clone().detach().requires_grad_(True)
outputs = model(nes_images)
loss = criterion(outputs, labels)
model.zero_grad()
loss.backward()
data_grad = nes_images.grad.data
# 更新动量 (batch_size, channels, height, width)
momentum = decay * momentum + data_grad / torch.sum(torch.abs(data_grad), dim=(1, 2, 3), keepdim=True)
# 计算带动量的符号梯度
sign_data_grad = momentum.sign()
# 更新对抗样本
perturbed_images = perturbed_images + alpha * sign_data_grad
perturbed_images = torch.clamp(perturbed_images, original_images - epsilon, original_images + epsilon)
perturbed_images = perturbed_images.detach().requires_grad_(True)
return perturbed_images
攻击效果
代码汇总
nifgsm.py
import torch
import torch.nn as nn
def NI_FGSM(model, criterion, original_images, labels, epsilon, num_iterations=10, decay=1):
"""
NI-FGSM (Nesterov Iterative Fast Gradient Sign Method)
参数:
- model: 要攻击的模型
- criterion: 损失函数
- original_images: 原始图像
- labels: 原始图像的标签
- epsilon: 最大扰动幅度
- num_iterations: 迭代次数
- decay: 动量衰减因子
"""
# alpha 每次迭代步长
alpha = epsilon / num_iterations
# 复制原始图像作为初始的对抗样本
perturbed_images = original_images.clone().detach().requires_grad_(True)
momentum = torch.zeros_like(original_images).detach().to(original_images.device)
for _ in range(num_iterations):
# 先沿先前累积梯度的方向进行跳跃
nes_images = perturbed_images + alpha * decay * momentum
nes_images = nes_images.clone().detach().requires_grad_(True)
outputs = model(nes_images)
loss = criterion(outputs, labels)
model.zero_grad()
loss.backward()
data_grad = nes_images.grad.data
# 更新动量 (batch_size, channels, height, width)
momentum = decay * momentum + data_grad / torch.sum(torch.abs(data_grad), dim=(1, 2, 3), keepdim=True)
# 计算带动量的符号梯度
sign_data_grad = momentum.sign()
# 更新对抗样本
perturbed_images = perturbed_images + alpha * sign_data_grad
perturbed_images = torch.clamp(perturbed_images, original_images - epsilon, original_images + epsilon)
perturbed_images = perturbed_images.detach().requires_grad_(True)
return perturbed_images
train.py
import torch
import torch.nn as nn
import torchvision
import torchvision.transforms as transforms
from models import ResNet18
# 数据预处理
transform_train = transforms.Compose([
transforms.RandomCrop(32, padding=4),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010))
])
transform_test = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010))
])
# 加载Cifar10训练集和测试集
trainset = torchvision.datasets.CIFAR10(root='./data', train=True, download=False, transform=transform_train)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=128, shuffle=True, num_workers=2)
testset = torchvision.datasets.CIFAR10(root='./data', train=False, download=False, transform=transform_test)
testloader = torch.utils.data.DataLoader(testset, batch_size=100, shuffle=False, num_workers=2)
# 定义设备(GPU或CPU)
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
# 初始化模型
model = ResNet18(num_classes=10)
model.to(device)
# 定义损失函数和优化器
criterion = nn.CrossEntropyLoss()
optimizer = torch.optim.Adam(model.parameters(), lr=0.01)
if __name__ == "__main__":
# 训练模型
for epoch in range(10): # 可以根据实际情况调整训练轮数
running_loss = 0.0
for i, data in enumerate(trainloader, 0):
inputs, labels = data[0].to(device), data[1].to(device)
optimizer.zero_grad()
outputs = model(inputs)
loss = criterion(outputs, labels)
loss.backward()
optimizer.step()
running_loss += loss.item()
if i % 100 == 99:
print(f'Epoch {epoch + 1}, Batch {i + 1}: Loss = {running_loss / 100}')
running_loss = 0.0
torch.save(model.state_dict(), f'weights/epoch_{epoch + 1}.pth')
print('Finished Training')
advtest.py
import torch
import torch.nn as nn
import torchvision
import torchvision.transforms as transforms
from models import *
from attacks import *
import ssl
import os
from PIL import Image
import matplotlib.pyplot as plt
ssl._create_default_https_context = ssl._create_unverified_context
# 定义数据预处理操作
transform = transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize((0.491, 0.482, 0.446), (0.247, 0.243, 0.261))])
# 加载CIFAR10测试集
testset = torchvision.datasets.CIFAR10(root='./data', train=False,
download=False, transform=transform)
testloader = torch.utils.data.DataLoader(testset, batch_size=128,
shuffle=False, num_workers=2)
# 定义设备(GPU优先,若可用)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model = ResNet18(num_classes=10).to(device)
criterion = nn.CrossEntropyLoss()
# 加载模型权重
weights_path = "weights/epoch_10.pth"
model.load_state_dict(torch.load(weights_path, map_location=device))
if __name__ == "__main__":
# 在测试集上进行FGSM攻击并评估准确率
model.eval() # 设置为评估模式
correct = 0
total = 0
epsilon = 16 / 255 # 可以调整扰动强度
for data in testloader:
original_images, labels = data[0].to(device), data[1].to(device)
original_images.requires_grad = True
attack_name = 'NI-FGSM'
if attack_name == 'FGSM':
perturbed_images = FGSM(model, criterion, original_images, labels, epsilon)
elif attack_name == 'BIM':
perturbed_images = BIM(model, criterion, original_images, labels, epsilon)
elif attack_name == 'MI-FGSM':
perturbed_images = MI_FGSM(model, criterion, original_images, labels, epsilon)
elif attack_name == 'NI-FGSM':
perturbed_images = NI_FGSM(model, criterion, original_images, labels, epsilon)
perturbed_outputs = model(perturbed_images)
_, predicted = torch.max(perturbed_outputs.data, 1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
accuracy = 100 * correct / total
# Attack Success Rate
ASR = 100 - accuracy
print(f'Load ResNet Model Weight from {weights_path}')
print(f'epsilon: {epsilon}')
print(f'ASR of {attack_name} : {ASR}%')