Composing Efficient, Robust Tests for Policy Selection

Autor:	Morrill, Dustin, Walsh, Thomas J., Hernandez, Daniel, Wurman, Peter R., Stone, Peter
Rok vydání:	2023
Předmět:	Computer Science - Machine Learning Computer Science - Artificial Intelligence Computer Science - Computer Science and Game Theory B.8.1 I.2.6
Druh dokumentu:	Working Paper
Popis:	Modern reinforcement learning systems produce many high-quality policies throughout the learning process. However, to choose which policy to actually deploy in the real world, they must be tested under an intractable number of environmental conditions. We introduce RPOSST, an algorithm to select a small set of test cases from a larger pool based on a relatively small number of sample evaluations. RPOSST treats the test case selection problem as a two-player game and optimizes a solution with provable $k$-of-$N$ robustness, bounding the error relative to a test that used all the test cases in the pool. Empirical results demonstrate that RPOSST finds a small set of test cases that identify high quality policies in a toy one-shot game, poker datasets, and a high-fidelity racing simulator. Comment: 26 pages, 13 figures. To appear in Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence (UAI 2023)
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2306.07372 Zobrazit plný text záznamu View this record from Arxiv