Research Repository

A Comparison of Self-Play Algorithms Under a Generalized Framework

Hernandez, Daniel and Denamganai, Kevin and Devlin, Sam and Samothrakis, Spyridon and Walker, James Alfred (2022) 'A Comparison of Self-Play Algorithms Under a Generalized Framework.' IEEE Transactions on Games, 14 (2). pp. 221-231. ISSN 2475-1502

2006.04471v1.pdf - Accepted Version

Download (1MB) | Preview


The notion of self-play, albeit often cited in multiagent Reinforcement Learning as a process by which to train agent policies from scratch, has received little efforts to be taxonomized within a formal model. We present a formalized framework, with clearly defined assumptions, which encapsulates the meaning of self-play as abstracted from various existing self-play algorithms. This framework is framed as an approximation to a theoretical solution concept for multiagent training. Through a novel qualitative visualization metric, on a simple environment, we show that different self-play algorithms generate different distributions of episode trajectories, leading to different explorations of the policy space by the learning agents. Quantitatively, on two environments, we analyze the learning dynamics of policies trained under different self-play algorithms captured under our framework and perform cross self-play performance comparisons. Our results indicate that, throughout training, various widely used self-play algorithms exhibit cyclic policy evolutions and that the choice of self-play algorithm significantly affects the final performance of trained agents.

Item Type: Article
Uncontrolled Keywords: Training; Games; Measurement; Statistics; Sociology; Heuristic algorithms; Reinforcement learning; Emergent phenomena; machine learning; multi-agent systems
Divisions: Faculty of Science and Health
Faculty of Science and Health > Computer Science and Electronic Engineering, School of
SWORD Depositor: Elements
Depositing User: Elements
Date Deposited: 13 Oct 2021 14:04
Last Modified: 23 Sep 2022 19:40

Actions (login required)

View Item View Item