Summary of Not All Prompts Are Secure: a Switchable Backdoor Attack Against Pre-trained Vision Transformers, by Sheng Yang et al.
Not All Prompts Are Secure: A Switchable Backdoor Attack Against Pre-trained Vision Transformers
by Sheng Yang, Jiawang Bai, Kuofeng Gao, Yong Yang, Yiming Li, Shu-tao Xia
First submitted to arxiv on: 17 May 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 The paper explores a novel security threat to vision transformers, specifically the pre-training and prompting paradigm. It proposes a backdoor attack that can turn a benign model into a malicious one by introducing an extra prompt token, called the switch token. This attack, named SWARM, learns a trigger and prompt tokens including a switch token to achieve high success rates (95%+) in various visual recognition tasks while remaining stealthy. The paper also introduces cross-mode feature distillation to reduce the effect of the switch token on clean samples. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper shows how to make backdoor attacks possible on pre-trained vision transformers, which can have serious consequences for users of cloud API. It proposes a new attack called SWARM that can be very hard to detect and remove. The attack is designed to be stealthy and can achieve high success rates in various visual recognition tasks. |
Keywords
» Artificial intelligence » Distillation » Prompt » Prompting » Token
