Metraipolis › archive › dc_f81ee23fa12e

Escaping Path Mirages in Offline Goal-Conditioned Reinforcement Learning — arXiv:2610.40130

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.40130
License (as archived): CC0-1.0
Archived: 2026-10-31T06:14:46.725Z
SHA-256 of the archived text: 0e15bda5020a3634a37a68185c328b7cd774a7e8d60af7ca2f77a45826f8a870
Stable URL of this page: https://metraipolis.com/documento/dc_f81ee23fa12e

How to cite

Escaping Path Mirages in Offline Goal-Conditioned Reinforcement Learning — arXiv:2610.40130 — arXiv (en), https://arxiv.org/abs/2610.40130 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f81ee23fa12e (Metraipolis archive dc_f81ee23fa12e, sha256 0e15bda5020a3634a37a68185c328b7cd774a7e8d60af7ca2f77a45826f8a870)

BibTeX

@misc{dcf81ee23fa12e,
  title = {Escaping Path Mirages in Offline Goal-Conditioned Reinforcement Learning — arXiv:2610.40130},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.40130},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f81ee23fa12e; sha256 0e15bda5020a3634a37a68185c328b7cd774a7e8d60af7ca2f77a45826f8a870},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.40130. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Offline goal-conditioned reinforcement learning remains challenging in stochastic long-horizon settings, where compounding value estimation errors hinder reliable goal reaching. While prior methods have sought to address this challenge through various approaches, a key challenge remains path mirage, where agents overcommit to spuriously successful trajectories in offline datasets induced by stochasticity, often leading to failure on difficult tasks. To address this issue, we propose Branch-aware Graph Planning (BGP), which captures stochastic branching structures by identifying high-variability points as graph nodes and designing goal-conditioned edges with segmented learning to avoid unreliable branches and reduce unnecessary stochasticity along paths. As a result, BGP enables reliable goal reaching even in highly stochastic long-horizon environments, and experiments on diverse OGBench tasks show that it substantially outperforms prior state-of-the-art offline goal-conditioned RL methods. Our implementation is available at this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.40130 — 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 2026-10-31T06:07:08.755Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).