AI RESEARCH
Self-Improving Skill Learning for Robust Skill-based Meta-Reinforcement Learning
arXiv CS.LG
•
ArXi:2502.03752v5 Announce Type: replace Meta-reinforcement learning (Meta-RL) facilitates rapid adaptation to unseen tasks but faces challenges in long-horizon environments. Skill-based approaches tackle this by decomposing state-action sequences into reusable skills and employing hierarchical decision-making. However, these methods are highly susceptible to noisy offline nstrations, leading to unstable skill learning and degraded performance.