Summary of Even-ifs From If-onlys: Are the Best Semi-factual Explanations Found Using Counterfactuals As Guides?, by Saugat Aryal et al.
Even-Ifs From If-Onlys: Are the Best Semi-Factual Explanations Found Using Counterfactuals As Guides?
by Saugat Aryal, Mark T. Keane
First submitted to arxiv on: 1 Mar 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- 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 Recently, “if-only” counterfactual explanations have gained popularity in Explainable AI (XAI) for describing feature-input changes resulting in altered decision-outcomes. A newer concept, semi-factual “even-if” explanations, highlight input changes that don’t affect the decision. This paper explores 8 semi-factual methods on 7 datasets using 5 metrics to investigate whether counterfactual guidance is necessary for producing high-quality explanations. The results suggest that computing other aspects of the decision space leads to better XAI outcomes, rather than relying on counterfactuals. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine trying to understand why a computer made a certain decision. “If-only” explanations try to figure out what changes would change the decision. A new way of explaining things is called “even-if”. This approach shows which inputs wouldn’t change the decision. Researchers tested 8 ways to do this on different datasets and found that it’s better to look at other aspects of how the computer made the decision rather than relying on these explanations. |
