Summary of Benchmarking Transferable Adversarial Attacks, by Zhibo Jin et al.
Benchmarking Transferable Adversarial Attacks
by Zhibo Jin, Jiayu Zhang, Zhiyu Zhu, Huaming Chen
First submitted to arxiv on: 1 Feb 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: 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 study provides an in-depth review of the transferability aspect of adversarial attacks on deep learning models. The authors systematically categorize and evaluate various methodologies designed to enhance transferability, including Generative Structure, Semantic Similarity, Gradient Editing, Target Modification, and Ensemble Approach. A benchmark framework called TAA-Bench is introduced, integrating ten leading methodologies for comparative analysis across different model architectures. The review examines the efficacy and constraints of each method, highlighting their underlying principles and practical applications. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Adversarial attacks on deep learning models are a big problem. This study takes a close look at how to make these attacks work across different types of models. They studied many ways to make this happen, including using fake data, looking for similar meanings, editing the attack to match the model, changing what the attack is targeting, and combining different approaches. The authors also created a special set of tools called TAA-Bench that lets researchers compare how well these different methods work. |
Keywords
* Artificial intelligence * Deep learning * Transferability
