Summary of Predictive Performance Comparison Of Decision Policies Under Confounding, by Luke Guerdan et al.
Predictive Performance Comparison of Decision Policies Under Confounding
by Luke Guerdan, Amanda Coston, Kenneth Holstein, Zhiwei Steven Wu
First submitted to arxiv on: 1 Apr 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computers and Society (cs.CY); 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 Machine learning models are often introduced to decision-making tasks, claiming to improve performance over existing policies. However, comparing predictive performance against an under-specified and data-dependent existing policy is challenging. To address this issue, we propose a method to compare the predictive performance of decision policies using modern identification approaches from causal inference and off-policy evaluation literatures. Our key insight is that certain regions of uncertainty can be safely ignored in policy comparisons. We develop a practical approach for finite-sample estimation of regret intervals under no assumptions on the parametric form of the status quo policy. We theoretically verify our framework and demonstrate its effectiveness using synthetic data experiments. Finally, we apply our framework to support a pre-deployment evaluation of a proposed modification to a healthcare enrollment policy. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Decision-making models can help improve decisions by predicting outcomes. But how do we compare these models to the way things are done now? It’s hard because the current approach is not well-defined and depends on many unknown factors. To solve this problem, researchers propose a new method for comparing model predictions with existing decision-making policies. The idea is that some uncertainties can be ignored when comparing the two. This new approach provides a practical way to estimate how much better or worse a new policy would perform compared to the current one. The authors test their method using fake data and also apply it to a real-world problem in healthcare. |
Keywords
* Artificial intelligence * Inference * Machine learning * Synthetic data
