Summary of Offline Imitation Learning by Controlling the Effective Planning Horizon, By Hee-jun Ahn et al.
Offline Imitation Learning by Controlling the Effective Planning Horizon
by Hee-Jun Ahn, Seong-Woong Shim, Byung-Jun Lee
First submitted to arxiv on: 18 Jan 2024
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 The paper investigates offline imitation learning, where a policy is learned from a limited number of expert demonstrations and an additional dataset of suboptimal behaviors. The authors focus on minimizing the divergence between state-action visitation distributions to consider future consequences. However, they find that existing algorithms suffer from magnified approximation errors when reducing the planning horizon, leading to performance degradation. Instead of explicit regularization, the authors propose controlling the effective planning horizon to improve imitation learning benchmarks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper explores how to make machines learn from a few expert examples and some not-so-good attempts. The goal is to teach the machine to do what the experts did. The problem is that the machine might get confused if it’s trying to predict what will happen in the future. The authors looked at two ways to fix this: one way uses math to make the predictions better, while the other way changes how the machine plans for the future. They found that the second method works better and can even beat some current methods. |
Keywords
* Artificial intelligence * Regularization
