Summary of Seabo: a Simple Search-based Method For Offline Imitation Learning, by Jiafei Lyu et al.
SEABO: A Simple Search-Based Method for Offline Imitation Learning
by Jiafei Lyu, Xiaoteng Ma, Le Wan, Runze Liu, Xiu Li, Zongqing Lu
First submitted to arxiv on: 6 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 In this paper, researchers tackle the challenge of offline reinforcement learning (RL) by proposing a new method for imitation learning. Offline RL learns from static datasets without interacting with the environment, but relies heavily on annotated reward labels. The proposed method, SEABO, uses an unsupervised search-based approach to learn a reward function based on expert data and unlabeled data. Experimental results show that SEABO achieves competitive performance to offline RL algorithms with ground-truth rewards and outperforms prior methods across many tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Offline learning is important for machines because it helps them learn from data without needing to interact with the environment. The problem is that we often need to create a reward function, which can be hard work. To solve this, researchers have created a new way to imitate expert behavior using only observations and no rewards. This method is called SEABO and it’s really good at learning from data. |
Keywords
* Artificial intelligence * Reinforcement learning * Unsupervised
