Metraipolis › archive › dc_3982c135bc3e
Disagreement-Regularized Imitation Learning for Image-Based Continuous Control with Gaussian and Beta Policies — arXiv:2609.38407
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.38407
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T13:45:15.710Z
- SHA-256 of the archived text
74492c20f9ecec1c1240c9770175bb173a0c332de24a22dce89a99a44f3fe279- Stable URL of this page
https://metraipolis.com/documento/dc_3982c135bc3e
How to cite
Disagreement-Regularized Imitation Learning for Image-Based Continuous Control with Gaussian and Beta Policies — arXiv:2609.38407 — arXiv (en), https://arxiv.org/abs/2609.38407 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_3982c135bc3e (Metraipolis archive dc_3982c135bc3e, sha256 74492c20f9ecec1c1240c9770175bb173a0c332de24a22dce89a99a44f3fe279)
BibTeX
@misc{dc3982c135bc3e,
title = {Disagreement-Regularized Imitation Learning for Image-Based Continuous Control with Gaussian and Beta Policies — arXiv:2609.38407},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.38407},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_3982c135bc3e; sha256 74492c20f9ecec1c1240c9770175bb173a0c332de24a22dce89a99a44f3fe279},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38407. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Purpose: Behavior cloning can accumulate errors when a learned controller visits states outside the demonstrated distribution. This study evaluates whether Disagreement-Regularized Imitation Learning (DRIL), which converts disagreement among cloned policies into a reinforcement-learning reward, improves image-based continuous control. Methods: A controlled CarRacing study combines Gaussian and Beta learner policies, demonstrations from either a clipped Gaussian expert or an intrinsically bounded Beta expert, one or 20 trajectories, deterministic and stochastic evaluation, and three retained stages: behavior cloning, the highest 10-episode training-score checkpoint, and the final DRIL checkpoint. The disagreement ensemble contains five Gaussian policies in every variant. Each retained policy is evaluated over 100 procedurally generated episodes. Results: Score-selected DRIL produced its largest gains in the few-demonstration setting, improving over the strongest behavior-cloning mean by 61% with clipped-action demonstrations and by 112% with bounded-action demonstrations. With 20 trajectories, the advantage of DRIL narrowed; in the bounded-action regime, Beta behavior cloning remained about 7% above the best DRIL checkpoint. The experiments also show that the informativeness of the disagreement reward changes with the learner representation and training stage. Conclusion: DRIL can substantially improve few-demonstration visual continuous control, while bounded Beta policies provide strong behavior-cloning performance when more demonstrations are available. The results highlight the joint importance of learner support,ensemble response, and checkpoint selection. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.LG). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.38407 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-07T13:18:10.200Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).