Enhancing Output Diversity Improves Conjugate Gradient-based Adversarial Attacks
Authors:
- Keiichiro YamamuraKYKeiichiro Yamamura
- Issa OeIOIssa Oe
- Hiroki IshikuraHIHiroki Ishikura
Preprint
Abstract
Deep neural networks are vulnerable to adversarial examples, and adversarial attacks that generate adversarial examples have been studied in this context. Existing studies imply that increasing the diversity of model outputs contributes to improving the attack performance. This study focuses on the Auto Conjugate Gradient (ACG) attack, which is inspired by the conjugate gradient method and has a high diversification performance. We hypothesized that increasing the distance between two consecutive search points would enhance the output diversity. To test our hypothesis, we propose Rescaling-ACG (ReACG), which automatically modifies the two components that significantly affect the distance between two consecutive search points, including the search direction and step size. ReACG showed higher attack performance than that of ACG, and is particularly effective for ImageNet models with several classification classes. Experimental results show that the distance between two consecutive search points enhances the output diversity and may help develop new potent attacks. The code is available at \url{https://github.com/yamamura-k/ReACG}Keywords
All Related Versions
This publication has 12 references indexed in Scilit:
- Improving the Accuracy-Robustness Trade-Off of Classifiers via Adaptive SmoothingSIAM Journal on Mathematics of Data Science, 2024
- A Comprehensive Study on Robustness of Image Classification Models: Benchmarking and RethinkingInternational Journal of Computer Vision, 2024
- A Light Recipe to Train Robust Vision Transformers2023
- Practical Evaluation of Adversarial Robustness via Adaptive Auto Attack2022
- Deep learning for object detection and scene perception in self-driving cars: Survey, challenges, and open issuesArray, 2021
- Composite Adversarial AttacksProceedings of the AAAI Conference on Artificial Intelligence, 2021
- Miss the Point: Targeted Adversarial Attack on Multiple Landmark Detection2020
- Towards Transferable Targeted Attack2020
- Audio Adversarial Examples: Targeted Attacks on Speech-to-Text2018
- Towards Evaluating the Robustness of Neural Networks2017