Summary of Oasis: Conditional Distribution Shaping For Offline Safe Reinforcement Learning, by Yihang Yao et al.
OASIS: Conditional Distribution Shaping for Offline Safe Reinforcement Learning
by Yihang Yao, Zhepeng Cen, Wenhao Ding, Haohong Lin, Shiqi Liu, Tingnan Zhang, Wenhao Yu, Ding Zhao
First submitted to arxiv on: 19 Jul 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 This paper proposes a novel approach to offline safe reinforcement learning (RL), which aims to train policies that satisfy constraints using pre-collected datasets. The current methods struggle with the mismatch between imperfect demonstrations and the desired safe and rewarding performance. OASIS, a new paradigm in offline safe RL, utilizes a conditional diffusion model to synthesize offline datasets, shaping the data distribution toward a beneficial target domain. This approach makes compliance with safety constraints through effective data utilization and regularization techniques, benefiting offline safe RL training. The comprehensive evaluations on public benchmarks and varying datasets showcase OASIS’s superiority in achieving high-reward behavior while satisfying the safety constraints, outperforming established baselines. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Offline reinforcement learning is a technique that helps train agents to make good decisions without needing direct feedback from humans. This paper introduces a new way of doing offline reinforcement learning called OASIS. OASIS uses computer vision and data manipulation to help the agent learn from imperfect demonstrations, which are like training examples for the agent. The goal is to get the agent to behave safely and rewardingly while following certain rules or constraints. |
Keywords
» Artificial intelligence » Diffusion model » Regularization » Reinforcement learning
