Summary of Pagar: Taming Reward Misalignment in Inverse Reinforcement Learning-based Imitation Learning with Protagonist Antagonist Guided Adversarial Reward, by Weichao Zhou et al.
PAGAR: Taming Reward Misalignment in Inverse Reinforcement Learning-Based Imitation Learning with Protagonist Antagonist Guided Adversarial Reward
by Weichao Zhou, Wenchao Li
First submitted to arxiv on: 2 Jun 2023
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 tackles the issue of imitation learning (IL) algorithms failing to achieve their intended task due to misalignment between the inferred reward function and the objective. The authors propose Protagonist Antagonist Guided Adversarial Reward (PAGAR), a semi-supervised reward design paradigm that trains policies under mixed reward functions, unlike traditional inverse reinforcement learning (IRL) approaches. PAGAR-based IL can avoid task failures caused by reward misalignment under specific theoretical conditions. The authors also present an on-and-off policy approach to implementing PAGAR and demonstrate its effectiveness in complex tasks and challenging transfer settings, outperforming standard IL baselines. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research paper solves a problem with how some machines learn from experts by doing things right or wrong. Sometimes, these machines don’t do what they’re supposed to because the reward system is not aligned correctly. The authors created a new way called PAGAR that trains machines to work well under different reward systems. This helps them avoid mistakes and do better in complex tasks. They tested this approach and found it works better than usual methods. |
Keywords
* Artificial intelligence * Reinforcement learning * Semi supervised
