中文

对抗鲁棒性的统一博弈论解释

机器学习 2021-11-09 v2 人工智能 计算机视觉与模式识别
作者: Jie Ren , Die Zhang , Yisen Wang , Lu Chen , Zhanpeng Zhou , Yiting Chen , Xu Cheng , Xin Wang , Meng Zhou , Jie Shi , Quanshi Zhang
在 arXiv 上查看 ↗ PDF ↗

摘要

本文提供了一种统一的视角来解释不同的对抗攻击与防御方法，即DNNs输入变量间的多阶交互视角。基于多阶交互，我们发现对抗攻击主要通过影响高阶交互来欺骗DNN。此外，我们发现对抗训练得到的DNN的鲁棒性来源于类别特定的低阶交互。我们的发现提供了一种统一对抗扰动与鲁棒性的潜在方法，能够以原理性的方式解释现有的防御方法。此外，我们的发现也修正了先前关于对抗学习特征的形状偏置的不准确理解。

关键词

adversarial training adversarial examples adversarial robustness

引用

@article{arxiv.2111.03536,
  title  = {A Unified Game-Theoretic Interpretation of Adversarial Robustness},
  author = {Jie Ren and Die Zhang and Yisen Wang and Lu Chen and Zhanpeng Zhou and Yiting Chen and Xu Cheng and Xin Wang and Meng Zhou and Jie Shi and Quanshi Zhang},
  journal= {arXiv preprint arXiv:2111.03536},
  year   = {2021}
}

备注

the previous version is arXiv:2103.07364, but I mistakenly apply a new ID for the paper

相关论文

查看更多相关论文 →
机器学习 · 计算机科学
A Unified Game-Theoretic Interpretation of Adversarial Robustness
Jie Ren, Die Zhang, Yisen Wang, Lu Chen +7
2021-11-10
机器学习 · 计算机科学
Bridging Adversarial Robustness and Gradient Interpretability
Beomsu Kim, Junghoon Seo, Taegyun Jeon
2019-04-22
机器学习 · 统计学
Understanding Adversarial Behavior of DNNs by Disentangling Non-Robust and Robust Components in Performance Metric
Yujun Shi, Benben Liao, Guangyong Chen, Yun Liu +2
2019-06-07
机器学习 · 计算机科学
Game Theory for Adversarial Attacks and Defenses
Shorya Sharma
2024-01-08
机器学习 · 计算机科学
Recent Advances in Understanding Adversarial Robustness of Deep Neural Networks
Tao Bai, Jinqi Luo, Jun Zhao
2020-11-04
机器学习 · 计算机科学
A Unified Approach to Interpreting and Boosting Adversarial Transferability
Xin Wang, Jie Ren, Shuyun Lin, Xiangming Zhu +2
2023-12-04
机器学习 · 计算机科学
Exploring Architectural Ingredients of Adversarially Robust Deep Neural Networks
Hanxun Huang, Yisen Wang, Sarah Monazam Erfani, Quanquan Gu +2
2022-01-25
机器学习 · 计算机科学
A Theoretical Perspective on Subnetwork Contributions to Adversarial Robustness
Jovon Craig, Josh Andle, Theodore S. Nowak, Salimeh Yasaei Sekeh
2023-07-11
机器学习 · 计算机科学
Understanding Adversarial Attacks on Observations in Deep Reinforcement Learning
You Qiaoben, Chengyang Ying, Xinning Zhou, Hang Su +2
2023-02-28
计算机视觉与模式识别 · 计算机科学
A Self-supervised Approach for Adversarial Robustness
Muzammal Naseer, Salman Khan, Munawar Hayat, Fahad Shahbaz Khan +1
2020-06-11
计算机视觉与模式识别 · 计算机科学
Interpreting and Improving Adversarial Robustness of Deep Neural Networks with Neuron Sensitivity
Chongzhi Zhang, Aishan Liu, Xianglong Liu, Yitao Xu +3
2020-12-15
机器学习 · 计算机科学
Efficient Defenses Against Adversarial Attacks
Valentina Zantedeschi, Maria-Irina Nicolae, Ambrish Rawat
2017-08-31
计算机视觉与模式识别 · 计算机科学
Deep Defense: Training DNNs with Improved Adversarial Robustness
Ziang Yan, Yiwen Guo, Changshui Zhang
2018-12-21
计算机视觉与模式识别 · 计算机科学
Stylized Adversarial Defense
Muzammal Naseer, Salman Khan, Munawar Hayat, Fahad Shahbaz Khan +1
2022-09-19
机器学习 · 计算机科学
Interpreting Attributions and Interactions of Adversarial Attacks
Xin Wang, Shuyun Lin, Hao Zhang, Yufei Zhu +1
2021-08-17
机器学习 · 计算机科学
An Adversarial Approach for Explaining the Predictions of Deep Neural Networks
Arash Rahnama, Andrew Tseng
2020-09-29
机器学习 · 计算机科学
Proper Network Interpretability Helps Adversarial Robustness in Classification
Akhilan Boopathy, Sijia Liu, Gaoyuan Zhang, Cynthia Liu +3
2020-10-23
机器学习 · 计算机科学
Improving Adversarial Robustness via Mutual Information Estimation
Dawei Zhou, Nannan Wang, Xinbo Gao, Bo Han +3
2022-07-26
计算机视觉与模式识别 · 计算机科学
On the Properties of Adversarially-Trained CNNs
Mattia Carletti, Matteo Terzi, Gian Antonio Susto
2022-03-18
机器学习 · 计算机科学
Not So Robust After All: Evaluating the Robustness of Deep Neural Networks to Unseen Adversarial Attacks
Roman Garaev, Bader Rasheed, Adil Khan
2023-08-15
机器学习 · 计算机科学
Does simple trump complex? Comparing strategies for adversarial robustness in DNNs
William Brooks, Marelie H. Davel, Coenraad Mouton
2025-08-26
机器学习 · 计算机科学
On the Effectiveness of Adversarial Training against Backdoor Attacks
Yinghua Gao, Dongxian Wu, Jingfeng Zhang, Guanhao Gan +3
2022-02-23
计算机视觉与模式识别 · 计算机科学
Adversarial alignment: Breaking the trade-off between the strength of an attack and its relevance to human perception
Drew Linsley, Pinyuan Feng, Thibaut Boissin, Alekh Karkada Ashok +3
2023-06-07
机器学习 · 计算机科学
Enhancing Adversarial Robustness through Multi-Objective Representation Learning
Sedjro Salomon Hotegni, Sebastian Peitz
2025-07-01
机器学习 · 计算机科学
Recent Advances in Adversarial Training for Adversarial Robustness
Tao Bai, Jinqi Luo, Jun Zhao, Bihan Wen +1
2021-04-22