Research map / Multi-agent systems and games / Cooperative multi-agent RL
A General Learning Framework for Open Ad Hoc Teamwork Using Graph-based Policy Learning
Arrasy Rahman, Ignacio Carlucho, Niklas Höpner, Stefano V. Albrecht
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add GNN action-value estimation (vs QL) (ablation) | Q-learning learner · FortAttack (evaluation) | average return: -3.51 -> 7.01 | Improved |
| add GNN action-value estimation (vs QL) (ablation) | Q-learning learner · Wolfpack (evaluation) | average return: 20.57 -> 8.88 | Hurt |
| add GNN action-value estimation (vs QL) (ablation) | Q-learning learner · LBF (evaluation) | average return: 1.41 -> 2.07 | Improved |
| add agent model without joint action value (QL to QL-AM) (ablation) | QL · Wolfpack (evaluation) | average return: 20.57 -> 14.24 | Hurt |
| add agent model to GNN (GNN-AM) (ablation) | GNN · Wolfpack (evaluation) | average return: 8.88 -> 30.87 | Improved |
| add agent model to GNN (GNN-AM) (ablation) | GNN · LBF (evaluation) | average return: 2.07 -> 1.80 | Hurt |
| add joint action-value decomposition (GPL-Q vs GNN-AM) (ablation) | GNN-AM · FortAttack (evaluation) | average return: 8.12 -> 14.20 | Improved |
| add joint action-value decomposition (GPL-Q vs GNN-AM) (ablation) | GNN-AM · Wolfpack (evaluation) | average return: 30.87 -> 36.36 | Improved |
| train joint value with soft policy iteration instead of Q-learning | GPL-Q · FortAttack (evaluation) | average return: 14.20 -> 16.82 | Improved |
| use joint-training MARL (MADDPG) instead of GPL | GPL-Q · Wolfpack (evaluation) | average return: 36.36 -> 19.20 | Hurt |
| use DGN instead of GPL | GPL-Q · LBF (evaluation) | average return: 2.32 -> 0.64 | Hurt |
Closest papers by meaning
- Open Ad Hoc Teamwork with Cooperative Game Theory (ICML 2024 · similarity 0.93)
- N-agent Ad Hoc Teamwork (NeurIPS 2024 · similarity 0.91)
- ROTATE: Regret-driven Open-ended Training for Ad Hoc Teamwork (NeurIPS 2026 · similarity 0.90)
- Ad Hoc Teamwork via Offline Goal-Based Decision Transformers (ICML 2025 · similarity 0.89)
- Minimum Coverage Sets for Training Robust Ad Hoc Teamwork Agents (AAAI 2024 · similarity 0.89)
- NestRL: A Nested Training Regime for Mutual Adaptation in Human–AI Teaming (NeurIPS 2026 · similarity 0.87)
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.
