{"ID":2860966,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.03013","arxiv_id":"2510.03013","title":"Distributional Inverse Reinforcement Learning","abstract":"We propose a distributional framework for offline Inverse Reinforcement Learning (IRL) that jointly models uncertainty over reward functions and full distributions of returns. Unlike conventional IRL approaches that recover a deterministic reward estimate or match only expected returns, our method captures richer structure in expert behavior, particularly in learning the reward distribution, by minimizing first-order stochastic dominance (FSD) violations and thus integrating distortion risk measures (DRMs) into policy learning, enabling the recovery of both reward distributions and distribution-aware policies. This formulation is well-suited for behavior analysis and risk-aware imitation learning. Theoretical analysis shows that the algorithm converges with $\\mathcal{O}(\\varepsilon^{-2})$ iteration complexity. Empirical results on synthetic benchmarks, real-world neurobehavioral data, and MuJoCo control tasks demonstrate that our method recovers expressive reward representations and achieves state-of-the-art performance.","short_abstract":"We propose a distributional framework for offline Inverse Reinforcement Learning (IRL) that jointly models uncertainty over reward functions and full distributions of returns. Unlike conventional IRL approaches that recover a deterministic reward estimate or match only expected returns, our method captures richer struc...","url_abs":"https://arxiv.org/abs/2510.03013","url_pdf":"https://arxiv.org/pdf/2510.03013v4","authors":"[\"Feiyang Wu\",\"Ye Zhao\",\"Anqi Wu\"]","published":"2025-10-03T13:58:09Z","proceeding":"cs.LG","tasks":"[\"cs.LG\"]","methods":"[\"Reinforcement Learning\"]","has_code":false}