Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Vadori, Nelson, Savani, Rahul
Format: Preprint
Veröffentlicht: 2023
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2306.05366
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866929265329569792
author Vadori, Nelson
Savani, Rahul
author_facet Vadori, Nelson
Savani, Rahul
contents It was recently observed that Elo ratings fail at preserving transitive relations among strategies and therefore cannot correctly extract the transitive component of a game. We provide a characterization of transitive games as a weak variant of ordinal potential games and show that Elo ratings actually do preserve transitivity when computed in the right space, using suitable invertible mappings. Leveraging this insight, we introduce a new game decomposition of an arbitrary game into transitive and cyclic components that is learnt using a neural network-based architecture and that prioritises capturing the sign pattern of the game, namely transitive and cyclic relations among strategies. We link our approach to the known concept of sign-rank, and evaluate our methodology using both toy examples and empirical data from real-world games.
format Preprint
id arxiv_https___arxiv_org_abs_2306_05366
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Ordinal Potential-based Player Rating
Vadori, Nelson
Savani, Rahul
Computer Science and Game Theory
Machine Learning
It was recently observed that Elo ratings fail at preserving transitive relations among strategies and therefore cannot correctly extract the transitive component of a game. We provide a characterization of transitive games as a weak variant of ordinal potential games and show that Elo ratings actually do preserve transitivity when computed in the right space, using suitable invertible mappings. Leveraging this insight, we introduce a new game decomposition of an arbitrary game into transitive and cyclic components that is learnt using a neural network-based architecture and that prioritises capturing the sign pattern of the game, namely transitive and cyclic relations among strategies. We link our approach to the known concept of sign-rank, and evaluate our methodology using both toy examples and empirical data from real-world games.
title Ordinal Potential-based Player Rating
topic Computer Science and Game Theory
Machine Learning
url https://arxiv.org/abs/2306.05366