Research map / Reinforcement learning / Policy gradient and actor-critic
Efficient Safe Meta-Reinforcement Learning: Provable Near-Optimality and Anytime Safety
Read the paperCheck your idea against 108,967 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 a safe meta-RL framework with one-step closed-form safe policy adaptation plus safe meta-policy training, aiming for near-optimal adaptation with anytime safety.
Problem
Safe meta-reinforcement learning: agents must adapt efficiently to unseen tasks while satisfying safety constraints at all times during adaptation.
Approach
A one-step safe policy adaptation method with a closed-form solution, paired with safe meta-policy training, to provide provable near-optimality and anytime safety during adaptation.
Closest papers by meaning
- Constrained Meta Reinforcement Learning with Provable Test-Time Safety (ICML 2026 · similarity 0.93)
- Constrained Meta-Reinforcement Learning for Adaptable Safety Guarantee with Differentiable Convex Programming (AAAI 2024 · similarity 0.92)
- Meta-Reinforcement Learning with Universal Policy Adaptation: Provable Near-Optimality under All-task Optimum Comparator (NeurIPS 2024 · similarity 0.91)
- Balance Reward and Safety Optimization for Safe Reinforcement Learning: A Perspective of Gradient Manipulation (AAAI 2024 · similarity 0.90)
- Sampling-Based Safe Reinforcement Learning (NeurIPS 2026 · similarity 0.90)
- Enhancing Efficiency of Safe Reinforcement Learning via Sample Manipulation (NeurIPS 2024 · similarity 0.90)
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.