Enhancing Multi-Agent Robustness: Addressing the Off-Diagonal Problem with Population-Based Training

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11320%2F24%3A10493454" target="_blank" >RIV/00216208:11320/24:10493454 - isvavai.cz</a>

Výsledek na webu

<a href="https://doi.org/10.1109/ICTAI62512.2024.00106" target="_blank" >https://doi.org/10.1109/ICTAI62512.2024.00106</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/ICTAI62512.2024.00106" target="_blank" >10.1109/ICTAI62512.2024.00106</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Enhancing Multi-Agent Robustness: Addressing the Off-Diagonal Problem with Population-Based Training

Popis výsledku v původním jazyce

Cooperation with previously unseen agents in multi-agent cooperative scenario is a challenging problem. Agents trained together tend to rely on certain conventions discovered during the training that may not be present when they should cooperate with new agents. This leads to unsatisfying performance in such cases. We call this the off-diagonal problem. In this paper, we investigate this problem on more than 20 maps from the Overcooked environment. First, we propose and evaluate a number of metrics to quantify the off-diagonal problem and then we propose a population-based training technique to alleviate this problem. The results show that the proposed metrics can be used to divide the maps into groups based on how much they are affected by this problem, and the population-based training improves the performance of the agents especially in those maps, where the problem is present, without having negative consequences in the other maps.

Název v anglickém jazyce

Enhancing Multi-Agent Robustness: Addressing the Off-Diagonal Problem with Population-Based Training

Popis výsledku anglicky

Cooperation with previously unseen agents in multi-agent cooperative scenario is a challenging problem. Agents trained together tend to rely on certain conventions discovered during the training that may not be present when they should cooperate with new agents. This leads to unsatisfying performance in such cases. We call this the off-diagonal problem. In this paper, we investigate this problem on more than 20 maps from the Overcooked environment. First, we propose and evaluate a number of metrics to quantify the off-diagonal problem and then we propose a population-based training technique to alleviate this problem. The results show that the proposed metrics can be used to divide the maps into groups based on how much they are affected by this problem, and the population-based training improves the performance of the agents especially in those maps, where the problem is present, without having negative consequences in the other maps.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach<br>I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

International Conference on Tools for Artificial Intelligence (ICTAI)

ISBN

—

ISSN

1082-3409

e-ISSN

2375-0197

Počet stran výsledku

8

Strana od-do

714-721

Název nakladatele

IEEE Computer Society

Místo vydání

Los Alamitos, CA

Místo konání akce

Herndon, VA, USA

Datum konání akce

28. 10. 2024

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—