Summary of The Edge-of-reach Problem in Offline Model-based Reinforcement Learning, by Anya Sims et al.
The Edge-of-Reach Problem in Offline Model-Based Reinforcement Learning
by Anya Sims, Cong Lu, Jakob Foerster, Yee Whye Teh
First submitted to arxiv on: 19 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 reinforcement learning, which trains agents from pre-collected datasets. While model-based methods offer a potential solution by training an approximate dynamics model, the prevailing theory assumes online RL in this model, attributing any performance gap to dynamics model errors. However, the authors show that existing model-based methods fail if the learned dynamics model is replaced with the true error-free dynamics. This highlights the edge-of-reach problem, where truncation of rollouts leads to value overestimation and complete performance collapse. The paper proposes Reach-Aware Value Learning (RAVL), a simple and robust method addressing this issue. Key concepts include offline reinforcement learning, model-based methods, edge-of-reach problem, and Reach-Aware Value Learning. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Offline reinforcement learning trains agents from pre-collected datasets. This helps with decision-making without needing real-time data. The paper looks at how well certain models do in this area. They found that some models don’t work well if they think about the situation perfectly. Instead, they get stuck and make bad choices. To fix this, the authors suggest a new way to learn called Reach-Aware Value Learning. |
Keywords
* Artificial intelligence * Reinforcement learning
