Visualizing the Evolution of Multi-agent Game-playing Behaviors

Faculty/Professorship: Information Visualisation 
Author(s): Agarwal, Shivam  ; Latif, Shahid ; Rothweiler, Aristide; Beck, Fabian  
Title of the compilation: EuroVis 2022 - Posters
Corporate Body: The Eurographics Association
Conference: 24th EG Conference on Visualization, 13-17 June 2022, Rome
Publisher Information: Geneve
Year of publication: 2022
Pages: 23-25
ISBN: 978-3-03868-185-4
Language(s): English
DOI: 10.2312/EVP.20221111
Analyzing the training evolution of AI agents in a multi-agent environment helps to understand changes in learned behaviors, as well as the sequence in which they are learned. We train an existing Pommerman team from scratch and, at regular intervals, let it battle against another top-performing team. We define thirteen game-specific behaviors and compute their occurrences in 600 matches. To investigate the evolution of these behaviors, we propose a visualization approach and showcase its usefulness in an application example.
GND Keywords: Mehragentensystem; Multivariate Daten; Spielverhalten; Künstliche Intelligenz; Visualisierung
Keywords: Multi-agent systems, multivariate data, evolving gameplay behaviors, AI training, visualization
DDC Classification: 004 Computer science  
RVK Classification: ST 274   
Peer Reviewed: Ja
International Distribution: Ja
Type: Conferenceobject
Release Date: 28. July 2022