Summary of Generating Likely Counterfactuals Using Sum-product Networks, by Jiri Nemecek et al.
Generating Likely Counterfactuals Using Sum-Product Networks
by Jiri Nemecek, Tomas Pevny, Jakub Marecek
First submitted to arxiv on: 25 Jan 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Machine Learning (cs.LG); Optimization and Control (math.OC)
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 proposes a system that provides high-likelihood explanations for AI decisions, meeting recent regulatory demands and user expectations. The approach focuses on finding counterfactuals that are close and sparse, while considering multiple criteria. To achieve this, the authors employ mixed-integer optimization (MIO) to model the search process. They also introduce an MIO formulation of a Sum-Product Network (SPN) to estimate the likelihood of a counterfactual. This innovation has independent interest beyond its application in explainable AI. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary AI systems make decisions that are often only explainable after they happen. To help with this, the paper develops a system that provides explanations for AI decisions. The goal is to find the best explanation by considering multiple factors. One important factor is how close the explanation is to what actually happened. Some previous methods have sacrificed closeness in order to make the explanations more plausible. This new approach uses something called mixed-integer optimization (MIO) to find the best explanation that meets many important criteria. |
Keywords
* Artificial intelligence * Likelihood * Optimization
