Authors
- Dustin Morrill
- Thomas Walsh
- Daniel Hernandez
- Peter R. Wurman
- Peter Stone
Venue
- UAI 2023
Date
- 2023
Share
Composing Efficient, Robust Tests for Policy Selection
Daniel Hernandez
Peter R. Wurman
UAI 2023
2023
Abstract
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.
Related Publications
The purpose of continual reinforcement learning is to train an agent on a sequence of tasks such that it learns the ones that appear later in the sequence while retaining theability to perform the tasks that appeared earlier. Experience replay is a popular method used to mak…
When designing reinforcement learning (RL) agents, a designer communicates the desired agent behavior through the definition of reward functions - numerical feedback given to the agent as reward or punishment for its actions. However, mapping desired behaviors to reward func…
Having explored an environment, intelligent agents should be able to transfer their knowledge to most downstream tasks within that environment. Referred to as ``zero-shot learning," this ability remains elusive for general-purpose reinforcement learning algorithms. While rec…
JOIN US
Shape the Future of AI with Sony AI
We want to hear from those of you who have a strong desire
to shape the future of AI.
