Metraipolis › archive › dc_735ced9f0834

Near-Optimal $\sqrt{K}$-Regret for Online Learning in Uninformed Markov Games — arXiv:2611.01649

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.01649
License (as archived)
CC0-1.0
Archived
2026-11-03T18:36:57.697Z
SHA-256 of the archived text
fa34528ac4be0c263c6093918693dd4e015dda647f311a7b03f8f22655d31c23
Stable URL of this page
https://metraipolis.com/documento/dc_735ced9f0834

How to cite

Near-Optimal $\sqrt{K}$-Regret for Online Learning in Uninformed Markov Games — arXiv:2611.01649 — arXiv (en), https://arxiv.org/abs/2611.01649 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_735ced9f0834 (Metraipolis archive dc_735ced9f0834, sha256 fa34528ac4be0c263c6093918693dd4e015dda647f311a7b03f8f22655d31c23)
BibTeX
@misc{dc735ced9f0834,
  title = {Near-Optimal $sqrtK$-Regret for Online Learning in Uninformed Markov Games — arXiv:2611.01649},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.01649},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_735ced9f0834; sha256 fa34528ac4be0c263c6093918693dd4e015dda647f311a7b03f8f22655d31c23},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.01649. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

We study online learning in two-player uninformed Markov games, where the opponent may follow an arbitrary, unknown policy in each episode and the learner does not observe the opponent's actions. [TWYS21] showed that achieving sublinear external regret requires an exponential dependence on the episode length $H$. Motivated by this limitation, they studied the weaker notion of Nash-value regret and obtained an $\widetilde{O}(K^{2/3})$ regret bound, where $K$ is the number of episodes. More recently, [LLZR26] introduced a stronger notion called empirical Nash-value regret and achieved an $\widetilde{O}(\sqrt{K})$ bound when the opponent's non-stationarity is moderate, including the case where the opponent follows a fixed policy. However, whether the $\sqrt{K}$ rate is achievable in the worst case remained open. In this paper, we resolve this open question by developing an algorithm that achieves $\widetilde{O}(H^{3/2}S\sqrt{AK})$ empirical Nash-value regret against arbitrary opponent policies, where $S$ is the number of states and $A$ is the number of learner's actions. Specifically, our algorithm combines global and local planning by maintaining a collection of candidate predictions of future losses, each associated with a policy. Global planning learns which candidates to favor across episodes, while local planning combines their policies differently at each state to make decisions. A key ingredient is to coordinate these two components to control the discrepancy between local loss estimates and the total loss of the executed policy, using only the learner's observed trajectories.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2611.01649 — 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 2026-11-03T18:12:54.512Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).