Counting Agents in Partially Observable Stochastic Games
Basic data for this talk
Type of talk: scientific talk
Name der Vortragenden: Karabulut, Nazlı Nur; Braun, Tanya
Date of talk: 25/09/2025
Talk language: English
Information about the event
Name of the event: ECSQARU-25 18th European Conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty
Event period: 23/09/2025 - 26/09/2025
Event location: Hagen
Event website: https://ecsqaru2025.krportal.org/
Abstract
Multi-agent decision-making under uncertainty can be modelled using partially observable stochastic games (POSGs), with numerous agents, partial observability, stochastic dynamics, and individual goals. However, POSGs are notoriously difficult to solve due to their exponential dependence on the number of agents. In this work, we present counting POSGs using the lifting technique of counting to compactly encode symmetries in a POSG, which enables using representative policies. We exploit the encoding for a counting version of the multi-agent dynamic programming operator to solve such a POSG. Doing so reduces the exponential dependence on the number of agents to a polynomial one, making the problem tractable with respect to agent numbers.Keywords: lifting; multi-agent decision making; POSG; histograms
Speakers from the University of Münster
| Karabulut, Nazli Nur | Junior professorship of practical computer science - modern aspects of data processing / data science (Prof. Braun) |