Improving Adversarial Transferability via Neuron Attribution-based Attacks
Jianping Zhang, Weibin Wu, Jen-tse Huang, Yizhan Huang, Wenxuan Wang, Yuxin Su, Michael R. Lyu
Abstract
Deep neural networks (DNNs) are known to be vulnerable to adversarial examples. It is thus imperative to devise effective attack algorithms to identify the deficiencies of DNNs beforehand in security-sensitive applications. To efficiently tackle the black-box setting where the target model's particulars are unknown, feature-level transfer-based attacks propose to contaminate the intermediate feature outputs of local models, and then directly employ the crafted adversarial samples to attack the target model. Due to the transferability of features, feature-level attacks have shown promise in synthesizing more transferable adversarial samples. However, existing feature-level attacks generally employ inaccurate neuron importance estimations, which deteriorates their transferability. To overcome such pitfalls, in this paper, we propose the Neuron Attribution-based Attack (NAA), which conducts feature-level attacks with more accurate neuron importance estimations. Specifically, we first completely attribute a model's output to each neuron in a middle layer. We then derive an approximation scheme of neuron attribution to tremendously reduce the computation overhead. Finally, we weight neurons based on their attribution results and launch feature-level attacks. Extensive experiments confirm the superiority of our approach to the state-of-the-art benchmarks. Our code is available at: https://github.com/jpzhang1810/NAA .
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
49 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- Improving Adversarial Transferability via Intermediate-level Perturbation Decay cites Nesterov Accelerated Gradient and Scale Invariance for Adversarial Attacks
- Improving Adversarial Transferability via Intermediate-level Perturbation Decay cites Feature Importance-aware Transferable Adversarial Attacks
- Nesterov Accelerated Gradient and Scale Invariance for Adversarial Attacks cites Towards Evaluating the Robustness of Neural Networks
- Fast is better than free: Revisiting adversarial training cites Towards Evaluating the Robustness of Neural Networks
- Feature Importance-aware Transferable Adversarial Attacks cites Towards Evaluating the Robustness of Neural Networks
- Feature Importance-aware Transferable Adversarial Attacks cites Nesterov Accelerated Gradient and Scale Invariance for Adversarial Attacks
- An Adaptive Model Ensemble Adversarial Attack for Boosting Adversarial Transferability cites Feature Importance-aware Transferable Adversarial Attacks
- Detecting Adversarial Data by Probing Multiple Perturbations Using Expected Perturbation Score cites Towards Evaluating the Robustness of Neural Networks
- Detecting Adversarial Data by Probing Multiple Perturbations Using Expected Perturbation Score cites Fast is better than free: Revisiting adversarial training
Cited by top-tier papers49
- An Adaptive Model Ensemble Adversarial Attack for Boosting Adversarial TransferabilityBin Chen, Jia-Li Yin, Shukai Chen, Bohao Chen et al.ICCV 2023 · 96 citations
- BiasAsker: Measuring the Bias in Conversational AI SystemYuxuan Wan, Wenxuan Wang, Pinjia He, Jiazhen Gu et al.FSE 2023 · 50 citations
- Improving Adversarial Transferability via Intermediate-level Perturbation DecayQizhang Li, Yiwen Guo, Wangmeng Zuo, Hao ChenNeurIPS 2023 · 43 citations
- Towards Reasonable Budget Allocation in Untargeted Graph Structure Attacks via Gradient DebiasZihan Liu, Yun Luo, Lirong Wu, Zicheng Liu et al.NeurIPS 2022 · 41 citations
- Detecting Adversarial Data by Probing Multiple Perturbations Using Expected Perturbation ScoreShuhai Zhang, Feng Liu, Jiahao Yang, Yifan Yang et al.ICML 2023 · 39 citations
Builds on9
- Towards Evaluating the Robustness of Neural NetworksNicholas Carlini, David A. WagnerS&P 2017 · 9,786 citations
- Fast is better than free: Revisiting adversarial trainingEric Wong, Leslie Rice, J. Zico KolterICLR 2020 · 1,352 citations
- Nesterov Accelerated Gradient and Scale Invariance for Adversarial AttacksJiadong Lin, Chuanbiao Song, Kun He, Liwei Wang et al.ICLR 2020 · 765 citations
- Do Adversarially Robust ImageNet Models Transfer Better?Hadi Salman, Andrew Ilyas, Logan Engstrom, Ashish Kapoor et al.NeurIPS 2020 · 506 citations
- Feature Importance-aware Transferable Adversarial AttacksZhibo Wang, Hengchang Guo, Zhifei Zhang, Wenxin Liu et al.ICCV 2021 · 306 citations
Related papers
- Pixel2Feature Attack (P2FA): Rethinking the Perturbed Space to Enhance Adversarial TransferabilityRenpu Liu, Hao Wu, Jiawei Zhang, Xin Cheng et al.ICML 2025
- Learning Transferable Adversarial PerturbationsKrishna Kanth Nakka, Mathieu SalzmannNeurIPS 2021 · 75 citations
- Improving the Transferability of Adversarial Samples With Adversarial TransformationsWeibin Wu, Yuxin Su, Michael R. Lyu, Irwin KingCVPR 2021
- Perturbing Across the Feature Hierarchy to Improve Standard and Strict Blackbox Attack TransferabilityNathan Inkawhich, Kevin J. Liang, Binghui Wang, Matthew Inkawhich et al.NeurIPS 2020 · 105 citations
- Enhancing Adversarial Example Transferability With an Intermediate Level AttackQian Huang, Isay Katsman, Zeqi Gu, Horace He et al.ICCV 2019 · 293 citations
