We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.GT

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computer Science and Game Theory

Title: Efficient Deviation Types and Learning for Hindsight Rationality in Extensive-Form Games

Abstract: Hindsight rationality is an approach to playing general-sum games that prescribes no-regret learning dynamics for individual agents with respect to a set of deviations, and further describes jointly rational behavior among multiple agents with mediated equilibria. To develop hindsight rational learning in sequential decision-making settings, we formalize behavioral deviations as a general class of deviations that respect the structure of extensive-form games. Integrating the idea of time selection into counterfactual regret minimization (CFR), we introduce the extensive-form regret minimization (EFR) algorithm that achieves hindsight rationality for any given set of behavioral deviations with computation that scales closely with the complexity of the set. We identify behavioral deviation subsets, the partial sequence deviation types, that subsume previously studied types and lead to efficient EFR instances in games with moderate lengths. In addition, we present a thorough empirical analysis of EFR instantiated with different deviation types in benchmark games, where we find that stronger types typically induce better performance.
Comments: Corrected technical report for the paper with the same title in the proceedings of the thirty-eighth International Conference on Machine Learning (ICML 2021), virtual. Compared to v5, this version removes the version indicator from an arXiv reference. 43 pages and 6 figures
Subjects: Computer Science and Game Theory (cs.GT); Artificial Intelligence (cs.AI)
Cite as: arXiv:2102.06973 [cs.GT]
  (or arXiv:2102.06973v7 [cs.GT] for this version)

Submission history

From: Dustin Morrill [view email]
[v1] Sat, 13 Feb 2021 18:12:53 GMT (1834kb,D)
[v2] Sat, 12 Jun 2021 03:42:10 GMT (3315kb,D)
[v3] Sat, 5 Mar 2022 21:06:53 GMT (3315kb,D)
[v4] Sun, 29 May 2022 16:54:40 GMT (3320kb,D)
[v5] Wed, 1 Jun 2022 23:28:39 GMT (3320kb,D)
[v6] Fri, 17 Jun 2022 22:16:20 GMT (3320kb,D)
[v7] Wed, 22 Jun 2022 23:16:01 GMT (3320kb,D)

Link back to: arXiv, form interface, contact.