Summary of Combining Experimental and Historical Data For Policy Evaluation, by Ting Li et al.
Combining Experimental and Historical Data for Policy Evaluation
by Ting Li, Chengchun Shi, Qianglin Wen, Yang Sui, Yongli Qin, Chunbo Lai, Hongtu Zhu
First submitted to arxiv on: 1 Jun 2024
Categories
- Main: Machine Learning (stat.ML)
- Secondary: Machine Learning (cs.LG); 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 investigates policy evaluation using multiple data sources in situations with one experimental dataset containing two arms, accompanied by a historical dataset generated under a single control arm. The authors propose novel integration methods that combine base policy value estimators from both datasets, optimized to minimize the mean square error (MSE) of the resulting estimator. They also apply the pessimistic principle to obtain more robust estimators and extend these developments to sequential decision making. The paper establishes non-asymptotic error bounds for the MSEs of the proposed estimators, demonstrating their oracle, efficiency, and robustness properties across various reward shift scenarios. Numerical experiments and real-data-based analyses from a ridesharing company demonstrate the superior performance of the proposed estimators. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research looks at how to evaluate policies using different types of data. They consider a situation where you have one set of data with two parts, and another set of data that happened in the past under one condition. The authors suggest new ways to combine these pieces of information to get a better estimate. They also make sure their methods are robust and can handle changing conditions. The paper shows that these methods work well in different scenarios and even works with real-world data from a ride-sharing company. |
Keywords
» Artificial intelligence » Mse
