Summary of Transcending Adversarial Perturbations: Manifold-aided Adversarial Examples with Legitimate Semantics, by Shuai Li et al.
Transcending Adversarial Perturbations: Manifold-Aided Adversarial Examples with Legitimate Semantics
by Shuai Li, Xiaoyu Jiang, Xiaoguang Ma
First submitted to arxiv on: 5 Feb 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
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 This paper addresses the issue of deep neural networks being vulnerable to adversarial examples. The authors propose a supervised semantic-transformation generative model that generates adversarial examples with real and legitimate semantics, allowing for continuous semantic variations between non-adversarial and adversarial examples. This approach leads to better visual quality, superior attack transferability, and more effective explanations of model vulnerabilities. The paper presents comprehensive experiments on MNIST and industrial defect datasets, demonstrating the potential of these generated adversarial examples as a generic tool. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A team of researchers found that deep neural networks are easily fooled by tiny changes in images. They wanted to make these attacks better and more reliable. To do this, they created a special kind of computer program that can change images in a way that makes them harder for the network to recognize. This new type of attack is very good at tricking the network and can even be used on real-world problems like finding defects in industrial products. The researchers tested their approach on two different datasets and found that it works really well. |
Keywords
* Artificial intelligence * Generative model * Semantics * Supervised * Transferability
