AI Paper Atlas

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

Boosting Multi-agent Reinforcement Learning via Contextual Prompting

Yue Deng, Zirui Wang, Xi Chen, Yin Zhang

JMLR 2023 · Multi-agent systems and games › Cooperative multi-agent RL · approach: Offline and networked 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
add contextual prompting (tree) to QMIXQMIX · Particle World Tag, 700k stepscumulative return: 56.05 ± 8.03 -> 153.83 ± 20.12Improved
add contextual prompting (tree) to IQLIQL · Particle World Tag, 700k stepscumulative return: 31.73 ± 10.86 -> 95.38 ± 18.37Improved
add contextual prompting (tree) to QMIXQMIX · Particle World Spread, 5k stepscumulative return: −160.66 ± 11.22 -> -156.18 ± 7.56Improved
add contextual prompting (tree) to QMIXQMIX · Particle World Reference, 5k stepscumulative return: −84.38 ± 9.92 -> -57.81 ± 5.73Improved
add contextual prompting (tree) to IQLIQL · Particle World Spread, 5k stepscumulative return: −214.48 ± 6.36 -> −173.11 ± 19.56Improved
add contextual prompting (tree) to QMIXQMIX · Particle World, averagecumulative return: −57.78 ± 11.71 -> −29.19 ± 10.59Improved
add contextual prompting (tree) to IQLIQL · Particle World, averagecumulative return: −78.53 ± 20.96 -> −41.61 ± 7.36Improved
use MCTS trajectory tree instead of replay-buffer trajectory pool for training the predictor (ablation)prompting QMIX · Particle World, averagecumulative return: −60.97 ± 11.81 -> −29.19 ± 10.59Improved
use trajectory tree instead of pool (ablation)prompting IQL · Particle World Tag, 700k stepscumulative return: 28.71 ± 10.18 -> 95.38 ± 18.37Improved
use trajectory pool prompting instead of plain IQL (ablation)IQL · Particle World Reference, 700k stepscumulative return: −50.81 ± 55.75 -> −38.45 ± 3.18Mixed
replace GRU contextual predictor with transformer predictor (ablation)prompting QMIX · Spread, Reference, Tagconvergence speed: about twice slowerHurt
replace GRU predictor with transformer predictor (ablation)prompting QMIX · TagreturnHurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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