Summary of Deep Adversarial Defense Against Multilevel-lp Attacks, by Ren Wang et al.
Deep Adversarial Defense Against Multilevel-Lp Attacks
by Ren Wang, Yuxuan Li, Alfred Hero
First submitted to arxiv on: 12 Jul 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Signal Processing (eess.SP)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper introduces the Efficient Robust Mode Connectivity (EMRC) method, a computationally efficient multilevel defense against multiple p-norm attacks. The approach blends two p-specific adversarially optimal models to provide good adversarial robustness for a range of p. Compared to traditional AT-, E-AT, and MSD methods, EMRC demonstrates better performance on various attacks and datasets/architectures, including CIFAR-10, CIFAR-100, PreResNet110, WideResNet, and ViT-Base. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper helps deep learning models become more secure by making them less vulnerable to bad data. It introduces a new way to make models robust against different types of attacks, which is better than existing methods. The approach uses two types of attack-resistant models and combines them in a smart way to provide good protection for a range of attacks. |
Keywords
* Artificial intelligence * Deep learning * Vit
