Summary of Feature Attribution with Necessity and Sufficiency Via Dual-stage Perturbation Test For Causal Explanation, by Xuexin Chen et al.
Feature Attribution with Necessity and Sufficiency via Dual-stage Perturbation Test for Causal Explanation
by Xuexin Chen, Ruichu Cai, Zhengting Huang, Yuxuan Zhu, Julien Horwood, Zhifeng Hao, Zijian Li, Jose Miguel Hernandez-Lobato
First submitted to arxiv on: 13 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Methodology (stat.ME)
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 tackles the issue of explainability in machine learning models, focusing on Feature Attribution Methods (FAMs) that evaluate feature importance through perturbation tests. Existing FAMs struggle to distinguish between features’ contributions when their prediction changes are similar after perturbation. To address this limitation, the authors introduce Feature Attribution with Necessity and Sufficiency (FANS), which identifies a neighborhood of input data where perturbing samples within this neighborhood has a high Probability of being Necessity and Sufficiency (PNS) cause for the change in predictions. FANS computes PNS via a heuristic strategy for estimating the neighborhood and a two-stage perturbation test involving factual and interventional reasoning. The authors demonstrate that FANS outperforms existing attribution methods on six benchmarks, making it a valuable tool for understanding the decision-making processes of machine learning models. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine you’re trying to figure out why a computer program made a certain decision. This paper helps make this process easier by introducing a new way to understand how machine learning models work. The authors are trying to solve a problem where existing methods can’t tell the difference between the importance of different features. They come up with a new approach called FANS, which looks at how changing small parts of the input data affects the model’s predictions. By using this approach, they show that it’s possible to get more accurate results than before and better understand why models make certain decisions. |
Keywords
* Artificial intelligence * Machine learning * Probability
