Summary of Addressing Key Challenges Of Adversarial Attacks and Defenses in the Tabular Domain: a Methodological Framework For Coherence and Consistency, by Yael Itzhakev et al.
Addressing Key Challenges of Adversarial Attacks and Defenses in the Tabular Domain: A Methodological Framework for Coherence and Consistency
by Yael Itzhakev, Amit Giloni, Yuval Elovici, Asaf Shabtai
First submitted to arxiv on: 10 Dec 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 A novel machine learning attack on tabular data models is proposed, which generates coherent samples to evade detection. The attack’s quality is evaluated using new criteria that assess its distinguishability from legitimate samples and feature consistency. SHAP explainability technique is utilized to identify inconsistencies in the model’s decision-making process caused by adversarial samples. This work paves the way for potential detection methods and improves established evaluation metrics for assessing attack quality. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Machine learning models are vulnerable to attacks, even when attackers only have access to the model’s outputs. Researchers want to know how well these attacks work. They evaluate attacks using success rate, perturbation magnitude, and query count. In this paper, we propose new ways to evaluate attacks on tabular data that generate coherent samples. We use SHAP explainability technique to see why models make mistakes when attacked. Our results show different attack strategies have strengths and weaknesses. |
Keywords
* Artificial intelligence * Machine learning
