AI Paper Atlas

Research map / Multi-agent systems and games / Cooperative multi-agent RL

Hierarchical Multi-Agent Skill Discovery

Mingyu Yang, Yaodong Yang, Zhenbo Lu, Wengang Zhou, Houqiang Li

NeurIPS 2023 · Multi-agent systems and games › Cooperative multi-agent RL · approach: Intrinsic-reward exploration for MARL

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use HMASD instead of EITI exploration bonusMARL baseline · Overcooked cramped_roomfinal episode reward: 0.122 -> 236.0Improved
use HMASD instead of EDTI exploration bonusMARL baseline · Overcooked coordination_ringfinal episode reward: 0.0 -> 80.0Improved
use HMASD instead of EITIMARL baseline · Overcooked asymmetric_advantagesfinal episode reward: 0.008 -> 295.1Improved
learn both team and individual skills (vs team skill only) (ablation)HMASD · SMAC 0-1 reward and OvercookedperformanceImproved
add extrinsic team reward to low-level reward (ablation)HMASD · SMAC 0-1 reward and OvercookedperformanceImproved
add intrinsic diversity rewards to low-level reward (ablation)HMASD · SMAC 0-1 reward and OvercookedperformanceImproved
add high-level skill coordinator (vs random skill assignment per episode) (ablation)HMASD · SMAC 0-1 reward and OvercookedperformanceImproved
vary skill interval k (ablation)HMASD · ablation Fig. 10performanceMixed
vary number of team and individual skills (ablation)HMASD · ablation Fig. 10performanceMixed
use HMASD instead of MAPPO and MATMAPPO, MAT · SMAC with 0-1 rewardwin rateImproved
use HMASD instead of MASERMASER · SMAC with 0-1 rewardwin rateImproved
increase state-observation space sizeHMASD · SMACuseful skills fraction: 12 of 50 skills useful, 24%Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy