Research map / Multi-agent systems and games / Cooperative multi-agent RL
Individual Reward Assisted Multi-Agent Reinforcement Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Learn separate individual and team policies with discrepancy constraints (IRAT) instead of a single policy on mixed rewards | MAPPO-based multi-agent policy gradient · Google Research Football 5-vs-5 half-court | average goal score | Helped |
| Learn separate individual and team policies with discrepancy constraints (IRAT) | MAPPO-based multi-agent policy gradient · Predator-Prey (MPE) | average team reward per episode | Helped |
| Learn separate individual and team policies with discrepancy constraints (IRAT) | MAPPO-based multi-agent policy gradient · Attack (MPE, conflicting individual reward) | average team reward per episode | Helped |
| Use individual policy with cooperation-oriented clip term (ICP) on individual policy (ablation) | IRAT · Predator-Prey (MPE) | average team reward per episode | Helped |
| Add team-policy importance-sampling-corrected clip term (TCP) (ablation) | IRAT · Predator-Prey (MPE) | average team reward per episode | Helped |
| Add KL distillation regularizer on individual policy (IKL) (ablation) | IRAT · Predator-Prey (MPE) | average team reward per episode | Helped |
| Remove KL regularizer on team policy (TKL) (ablation) | IRAT · Predator-Prey (MPE) | average team reward per episode | No clear effect |
| Reward shaping (summing individual and team rewards) | baseline · Predator-Prey (MPE) | average team reward per episode | Hurt |
| Use individual rewards only (no team reward) vs team reward only | baseline · Predator-Prey (MPE) | average team reward per episode | Helped |
| Individual rewards in football: conflict leads IR to degrade over training | Individual-Reward-Only baseline · Google Research Football 5-vs-5 half-court | average goal score | Hurt |
Closest papers by meaning
- LIIR: Learning Individual Intrinsic Reward in Multi-Agent Reinforcement Learning (NeurIPS 2019 · similarity 0.89)
- Learning to Incentivize Other Learning Agents (NeurIPS 2020 · similarity 0.89)
- Mutual-Information Regularized Multi-Agent Policy Iteration (NeurIPS 2023 · similarity 0.89)
- DIFFER:Decomposing Individual Reward for Fair Experience Replay in Multi-Agent Reinforcement Learning (NeurIPS 2023 · similarity 0.88)
- Learning to Collaborate with Unknown Agents in the Absence of Reward (AAAI 2025 · similarity 0.87)
- Intrinsic Action Tendency Consistency for Cooperative Multi-Agent Reinforcement Learning (AAAI 2024 · similarity 0.85)
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.