Madry A, Makelov A, Schmidt L, et al. Towards Deep Learning Models Resistant to Adversarial Attacks.[J]. arXiv: Machine Learning, 2017.

@article{madry2017towards,

title={Towards Deep Learning Models Resistant to Adversarial Attacks.},

author={Madry, Aleksander and Makelov, Aleksandar and Schmidt, Ludwig and Tsipras, Dimitris and Vladu, Adrian},

journal={arXiv: Machine Learning},

year={2017}}

利用特定的方法产生"坏"样本(Adversarial samples), 以此来促进网络的稳定性是当下的热点之一, 本文以实验为主, 比较PGD( projected gradient descent) 和 FGSM(fast gradient sign method)在不同数据下的表现, 以及由普通样本产生"坏"样本会出现的一些现象.

主要内容

Adversarial attacks 主要聚焦于下列问题:

\[\tag{2.1}
\min_{\theta} \rho (\theta) \quad where \quad \rho(\theta) =\mathbb{E}_{(x,y)\sim D}[\max_{\delta \in S} L(\theta, x+\delta, y)].
\]

其中\(S\)是我们指定的摄动集合, 直接一点就是\(|\delta| <constant\)之类.

通过FGSM产生"坏"样本:

\[x + \epsilon \: \mathrm{sgn}(\nabla_x L(\theta,x,y)).
\]

这个思想是很直接的(从线性感知器谈起, 具体看here).

PGD的思路是, 给定摄动集\(S\), 比如小于某个常数的摄动(e.g. \(\{\tilde{x}:\|x-\tilde{x}\|_{\infty}<c\}\)), 多次迭代寻找合适的adversarial samples:

\[x^{t+1} = \prod_{x+S} (x^t + \alpha \: \mathrm{sgn} (\nabla_x L(\theta,x, y)),
\]

其中\(\prod\)表示投影算子, 假设\(S=\{\tilde{x}:\|x-\tilde{x}\|_{\infty}<c\}\),

\[x^{t+1} = \arg \min_{z \in x+S} \frac{1}{2} \|z - (x^t + \alpha \: \mathrm{sgn} (\nabla_x L(\theta,x, y))\|_2^2,
\]

实际上, 可以分开讨论第\((i,j)\)个元素, \(y:=(x^t + \alpha \: \mathrm{sgn} (\nabla_x L(\theta,x, y))\), 只需找到\(z_{ij}\)使得

\[\|z_{ij}-y_{ij}\|_2
\]

最小即可. 此时有显示解为:

\[z_{ij}=
\left \{
\begin{array}{ll}
x_{ij} +c & y_{ij} > x_{ij}+c \\
x_{ij} -c & y_{ij} < x_{ij}-c \\
y_{ij} & else.
\end{array} \right.
\]

简而言之就是一个截断.

重复几次, 至到\(x^t\)被判断的类别与初始的\(x\)不同或者达到最大迭代次数.

Note

  • 如果我们训练网络能够免疫PGD的攻击, 那么其也能很大一部分其它的攻击.
  • FGSM对抗训练不能提高网络的稳定性(在摄动较大的时候).
  • weak models may fail to learn non-trival classfiers.
  • 网络越强(参数等程度)训练出来的稳定性越好, 同时可转移(指adversarial samples 在多个网络中被误判)会变差.

Towards Deep Learning Models Resistant to Adversarial Attacks的更多相关文章

  1. How to Grid Search Hyperparameters for Deep Learning Models in Python With Keras

    Hyperparameter optimization is a big part of deep learning. The reason is that neural networks are n ...

  2. a Javascript library for training Deep Learning models

    w强化算法和数学,来迎接机器学习.神经网络. http://cs.stanford.edu/people/karpathy/convnetjs/ ConvNetJS is a Javascript l ...

  3. Run Your Tensorflow Deep Learning Models on Google AI

    People commonly tend to put much effort on hyperparameter tuning and training while using Tensoflow& ...

  4. What are some good books/papers for learning deep learning?

    What's the most effective way to get started with deep learning?       29 Answers     Yoshua Bengio, ...

  5. (转) Awesome Deep Learning

    Awesome Deep Learning  Table of Contents Free Online Books Courses Videos and Lectures Papers Tutori ...

  6. (转)分布式深度学习系统构建 简介 Distributed Deep Learning

    HOME ABOUT CONTACT SUBSCRIBE VIA RSS   DEEP LEARNING FOR ENTERPRISE Distributed Deep Learning, Part ...

  7. The Brain vs Deep Learning Part I: Computational Complexity — Or Why the Singularity Is Nowhere Near

    The Brain vs Deep Learning Part I: Computational Complexity — Or Why the Singularity Is Nowhere Near ...

  8. Paper Reading——LEMNA:Explaining Deep Learning based Security Applications

    Motivation: The lack of transparency of the deep  learning models creates key barriers to establishi ...

  9. Coursera Deep Learning 2 Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization - week1, Assignment(Regularization)

    声明:所有内容来自coursera,作为个人学习笔记记录在这里. Regularization Welcome to the second assignment of this week. Deep ...

随机推荐

  1. A Child's History of England.30

    CHAPTER 10 ENGLAND UNDER HENRY THE FIRST, CALLED FINE-SCHOLAR Fine-scholar, on hearing of the Red Ki ...

  2. [php安全]原生类的利用

    php原生类的利用 查看原生类中具有魔法函数的类 $classes = get_declared_classes(); foreach ($classes as $class) { $methods ...

  3. Output of C++ Program | Set 16

    Predict the output of following C++ programs. Question 1 1 #include<iostream> 2 using namespac ...

  4. Linux学习 - 正则表达式

    一.正则表达式与通配符 正则表达式:在文件中匹配符合条件的字符串,正则是包含匹配 通配符:用来匹配符合条件的文件名,通配符是完全匹配 二.基础正则表达式 元字符 作用 a* a有0个或任意多个 . 除 ...

  5. PhoneGap本地将html打包成安卓App

    PhoneGap的在线打包有大小限制,超过30M的包无法在线打包.当然,可以把包里面的图片.声音文件去掉,然后打包.下载以后,解包,重新打包并签名.蛮麻烦的. 本地打包的简单方法如下: 下载安装Jav ...

  6. treeTable实现排序

    /* * * TreeTable 0.1 - Client-side TreeTable Viewer! * @requires jQuery v1.3 * * Dual licensed under ...

  7. Element-ui 中对表单进行验证

    Element-ui 中对表单(Form)绑定的对象中的对象属性进行校验 如果是直接绑定属性,是可以的,但是绑定对象中的属性就需要特别处理,需要在rules中添加双引号 " "或者 ...

  8. 【编程思想】【设计模式】【结构模式Structural】front_controller

    Python版 https://github.com/faif/python-patterns/blob/master/structural/front_controller.py #!/usr/bi ...

  9. 配置yum代理

    一.说明 很多内网环境无法使用yum 二.配置 1.安装nginx 2.配置 server { listen 808; #禁用multipart range功能 max_ranges 1; serve ...

  10. 万字长文入门 Redis 命令、事务、锁、订阅、性能测试

    作者:痴者工良 Redis 基本数据类型 Redis 中,常用的数据类型有以下几种: String:字符串类型,二进制安全字符串: Hash:哈希表: List 列表:链表结构,按照插入顺序排序的字符 ...