Research map / Reinforcement learning / Offline RL
Fat-to-Thin Policy Optimization: Offline Reinforcement Learning with Sparse Policies
Lingwei Zhu, Han Wang, Yukie Nagai
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes Fat-to-Thin Policy Optimization, an offline RL method that learns sparse continuous policies, which can assign strictly zero probability to some actions.
Problem
Offline reinforcement learning with continuous actions typically relies on Gaussian policies, which cannot represent sparse action distributions that place strictly zero probability on some actions, a property needed in safety-critical settings such as medicine.
Approach
Learns completely fat-free (sparse) continuous policies offline by combining sparse policy distributions with offline RL, moving from a dense ("fat") policy toward a sparse ("thin") one.
Closest papers by meaning
- Preference-based Policy Optimization from Sparse-reward Offline Dataset (ICLR 2026 · similarity 0.90)
- Generative OOD-regularized Model-based Policy Optimization (NeurIPS 2026 · similarity 0.89)
- MOPO: Model-based Offline Policy Optimization (NeurIPS 2020 · similarity 0.88)
- Offline Reinforcement Learning with Closed-Form Policy Improvement Operators (ICML 2023 · similarity 0.88)
- Relaxed Stationary Distribution Correction Estimation for Improved Offline Policy Optimization (AAAI 2024 · similarity 0.87)
- Off-Policy Primal-Dual Safe Reinforcement Learning (ICLR 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
