Research map / Reinforcement learning / Policy gradient and actor-critic
Improving Model-Based Reinforcement Learning by Converging to Flatter Minima
Shrinivas Ramasubramanian, Benjamin Freed, Alexandre Capone, Jeff Schneider
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.
Integrates Sharpness-Aware Minimization into world-model training in model-based RL so that the learned dynamics model converges to flatter minima, yielding better downstream policies.
Problem
Model-based RL depends on a learned dynamics model whose errors compound along imagined rollouts, degrading policy learning.
Approach
Encourage flatness in the dynamics model's training loss by plugging Sharpness-Aware Minimization in as a drop-in optimizer object for world-model training, so optimization steers toward flatter minima and produces a better policy.
Experimental setup
Builds on: Sharpness-Aware Minimization
Closest papers by meaning
- Towards General-Purpose Model-Free Reinforcement Learning (ICLR 2025 · similarity 0.88)
- PWM: Policy Learning with Multi-Task World Models (ICLR 2025 · similarity 0.88)
- The Edge-of-Reach Problem in Offline Model-Based Reinforcement Learning (NeurIPS 2024 · similarity 0.88)
- Trajectory Bellman Residual Minimization: A Simple Value-Based Method for LLM Reasoning (NeurIPS 2025 · similarity 0.87)
- Reward Model Learning vs. Direct Policy Optimization: A Comparative Analysis of Learning from Human Preferences (ICML 2024 · similarity 0.87)
- Policy-shaped prediction: avoiding distractions in model-based reinforcement learning (NeurIPS 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.
