Papers › Disentangling Sources of Risk for Distributional Multi-Agent Reinforcement Learning

Disentangling Sources of Risk for Distributional Multi-Agent Reinforcement Learning

29 Sep 2021archive 2025-07-28

Kyunghwan Son, Junsu Kim, Yung Yi, Jinwoo Shin

In cooperative multi-agent reinforcement learning, state transitions, rewards, and actions can all induce randomness (or uncertainty) in the observed long-term returns. These randomnesses are reflected from two risk sources: (a) agent-wise risk (i.e., how cooperative our teammates act for a given agent) and (b) environment-wise risk (i.e., transition stochasticity). Although these two sources are both important factors for learning robust policies of agents, prior works do not separate them or deal with only a single risk source, which could lead to suboptimal equilibria. In this paper, we propose Disentangled RIsk-sensitive Multi-Agent reinforcement learning (DRIMA), a novel framework being capable of disentangling risk sources. Our main idea is to separate risk-level leverages (i.e., quantiles) in both centralized training and decentralized execution with a hierarchical quantile structure and quantile regression. Our experiments demonstrate that DRIMA significantly outperforms prior-arts across various scenarios in StarCraft Multi-agent Challenge. Notably, DRIMA shows robust performance regardless of reward shaping, exploration schedule, where prior methods learn only a suboptimal policy.

PaperPDF

Code

No code repository is listed for this paper in the archive or in Syntology's graph.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Multi-agent Reinforcement LearningReinforcement LearningReinforcement Learning (RL)SMAC+Starcraftquantile regressionreinforcement-learning

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
SMAC+ Def_Armored_parallel DRIMA Median Win Rate 60.0 #3 of 10 Archive leaderboard report
SMAC+ Def_Armored_sequential DRIMA Median Win Rate 100 #1 of 11 Archive leaderboard report
SMAC+ Def_Infantry_parallel DRIMA Median Win Rate 100.0 #2 of 10 Archive leaderboard report
SMAC+ Def_Infantry_sequential DRIMA Median Win Rate 100 #2 of 11 Archive leaderboard report
SMAC+ Def_Outnumbered_parallel DRIMA Median Win Rate 70.0 #1 of 10 Archive leaderboard report
SMAC+ Def_Outnumbered_sequential DRIMA Median Win Rate 100 #1 of 11 Archive leaderboard report
SMAC+ Off_Complicated_parallel DRIMA Median Win Rate 100 #1 of 10 Archive leaderboard report
SMAC+ Off_Complicated_sequential DRIMA Median Win Rate 96.9 #1 of 4 Archive leaderboard report
SMAC+ Off_Distant_parallel DRIMA Median Win Rate 95.0 #1 of 10 Archive leaderboard report
SMAC+ Off_Distant_sequential DRIMA Median Win Rate 100 #1 of 4 Archive leaderboard report
SMAC+ Off_Hard_parallel DRIMA Median Win Rate 80.0 #1 of 10 Archive leaderboard report
SMAC+ Off_Hard_sequential DRIMA Median Win Rate 93.8 #2 of 4 Archive leaderboard report
SMAC+ Off_Near_parallel DRIMA Median Win Rate 95.0 #2 of 10 Archive leaderboard report
SMAC+ Off_Near_sequential DRIMA Median Win Rate 93.8 #1 of 4 Archive leaderboard report
SMAC+ Off_Superhard_parallel DRIMA Median Win Rate 0.0 #10 of 10 Archive leaderboard report
SMAC+ Off_Superhard_sequential DRIMA Median Win Rate 15.6 #1 of 4 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections