New Study Establishes Near-Minimax Sample Complexity for Multi-Objective Reinforcement Learning
A new paper addresses multi-objective reinforcement learning where preferences are specified only after data collection, requiring a single dataset to support a continuum of reward functions. It establishes near-minimax sample complexity bounds under generative models.
Read the full story at arXiv stat.ML ↗
Timeline · 1 report
- 2026-11-05 05:00 · arXiv stat.ML
Near-Minimax Sample Complexity for Multi-Objective Reinforcement Learning under Generative Models