Summary of Cosbo: Conservative Offline Simulation-based Policy Optimization, by Eshagh Kargar and Ville Kyrki
COSBO: Conservative Offline Simulation-Based Policy Optimization
by Eshagh Kargar, Ville Kyrki
First submitted to arxiv on: 22 Sep 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Robotics (cs.RO)
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 proposed method combines imperfect simulation environments with target environment data to train an offline reinforcement learning policy. This approach outperforms state-of-the-art methods CQL, MOPO, and COMBO in scenarios with diverse and challenging dynamics. The results demonstrate robust behavior across various experimental conditions. |
Low | GrooveSquid.com (original content) | Low Difficulty Summary Offline reinforcement learning trains models using live deployment data, but only allows choosing the best combination of behaviors present in the training data. Simulation environments can be used instead, but are limited by the simulation-to-reality gap. To combine the benefits of both approaches, a method combines simulator-generated data with target environment data to train an offline policy. The proposed method outperforms existing methods in scenarios with diverse dynamics and demonstrates robust behavior. |
Keywords
* Artificial intelligence * Reinforcement learning