Summary of Model Mimic Attack: Knowledge Distillation For Provably Transferable Adversarial Examples, by Kirill Lukyanov et al.
Model Mimic Attack: Knowledge Distillation for Provably Transferable Adversarial Examples
by Kirill Lukyanov, Andrew Perminov, Denis Turdakov, Mikhail Pautov
First submitted to arxiv on: 21 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 proposed approach utilizes knowledge distillation as a transfer-based black-box adversarial attack method, which iteratively trains a surrogate model on an expanding dataset. This technique provides provable guarantees on the success of the attack on classification neural networks, ensuring that if the student model has sufficient learning capabilities, it will find an adversarial example within a finite number of distillation iterations. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper explores how to create fake examples to fool artificial intelligence (AI) by using something called knowledge distillation. This is a new way to make AI models learn from each other. The researchers show that this method can be used to find bad inputs that will trick an AI model into making mistakes. They prove that if the student AI model is good enough, it will eventually figure out how to create fake examples that will fool the teacher AI model. |
Keywords
» Artificial intelligence » Classification » Distillation » Knowledge distillation » Student model
