AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

From Few to More: Large-Scale Dynamic Multiagent Curriculum Learning

Weixun Wang, Tianpei Yang, Yong Liu, Jianye Hao, Xiaotian Hao, Yujing Hu, Yingfeng Chen, Changjie Fan, Yang Gao

AAAI 2020 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: World-model-based agents · 119 citations (OpenAlex)

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
Add DyMA-CL with Max aggregationIQL · MAgent 50 vs 50Survivors: 0.54$\pm$2.65 -> 20.35$\pm$4.87Improved
Add DyMA-CL with Max aggregationIQL · MAgent 50 vs 50Kill count: 31.83$\pm$8.52 -> 50$\pm$0Improved
Add DyMA-CL with Sum aggregationPPO · MAgent 50 vs 50Kill count: 1.06$\pm$1.12 -> 38.25$\pm$7.33Improved
Add DyMA-CL with Max aggregationA2C · MAgent 50 vs 50Survivors: 3.65$\pm$5.58 -> 16.5$\pm$10.92Improved
Add DyMA-CL with Max aggregationA2C · MAgent 50 vs 50Kill count: 36.4$\pm$13.79 -> 47.71$\pm$5.61Improved
Add DyMA-CL with Mean aggregationACER · MAgent 50 vs 50Kill count: 4.84$\pm$2.8 -> 12.62$\pm$3.68Improved
Use DyMA-CL with model reload instead of learning from scratchIQL and VDN · StarCraft II 10I and 15Iaverage win rateImproved
Train from scratch rather than use DyMA-CLIQL · StarCraft II 15Iaverage win rate: 7.5e+6 steps; 0Hurt
Use SUM aggregation rather than MAX or MEAN (ablation)DyAN with IQL and VDN · StarCraft II 5I, 10I, and 15Iaverage win rateImproved
Replace the GNN aggregation operator with MAX (ablation)DyAN · MAgent 50 vs 50Survivors: IQL: 2.33$\pm$2.9 (Mean) vs 20.35$\pm$4.87 (Max); A2C: 6.8$\pm$7.39 (Mean) vs 16.5$\pm$10.92 (Max)Mixed
Use curriculum distillation instead of learning from scratchIQL and VDN · StarCraft II 10I and 15Iaverage win rateImproved
Reuse replay buffers from earlier curriculaIQL and VDN · StarCraft II 10I and 15Iaverage win rateImproved

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