Research map / Multi-agent systems and games / Cooperative multi-agent RL
Iterated Reasoning with Mutual Information in Cooperative and Byzantine Decentralized Teaming
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 |
|---|---|---|---|
| Use InfoPG instead of MOA | decentralized cooperative MARL · Co-op Pong | team reward: -0.3 -> 0.127 | Helped |
| Use Adv. InfoPG instead of MOA | decentralized cooperative MARL · Co-op Pong | team reward: -0.3 -> 0.25 | Helped |
| Use InfoPG instead of MOA | decentralized cooperative MARL · Pistonball | team reward: 4.10 -> 7.47 | Helped |
| Use Adv. InfoPG instead of MOA | decentralized cooperative MARL · Pistonball | team reward: 4.10 -> 7.33 | Helped |
| Use InfoPG instead of MOA | decentralized cooperative MARL · Multiwalker | team reward: 0.66 -> 3.56 | Helped |
| Use Adv. InfoPG instead of MOA | decentralized cooperative MARL · Multiwalker | team reward: 0.66 -> 11.81 | Helped |
| Use InfoPG instead of MOA | decentralized cooperative MARL · StarCraft II 3M | team reward: 2.78 -> 4.40 | Helped |
| Use Adv. InfoPG instead of MOA | decentralized cooperative MARL · StarCraft II 3M | team reward: 2.78 -> 3.73 | Helped |
| Increase reasoning depth from one step to k=2 or k=3 (ablation) | InfoPG · Co-op Pong | team reward | No clear effect |
| Increase reasoning depth from one step to k=2 or k=3 (ablation) | InfoPG · Pistonball and Multiwalker | team reward | Helped |
| Replace InfoPG with Adv. InfoPG when a random fraudulent agent is present (ablation) | Pistonball team · Pistonball fraudulent-agent experiment | team reward | Helped |
| Maximize MI without adapting to negative advantage in the presence of a random fraudulent agent (ablation) | InfoPG and MOA · Pistonball fraudulent-agent experiment | team reward | Hurt |
Closest papers by meaning
- Learning to Share in Networked Multi-Agent Reinforcement Learning (NeurIPS 2022 · similarity 0.89)
- Fully Decentralized Policies for Multi-Agent Systems: An Information Theoretic Approach (NeurIPS 2017 · similarity 0.89)
- Multi-agent cooperation through learning-aware policy gradients (ICLR 2025 · similarity 0.89)
- Mutual-Information Regularized Multi-Agent Policy Iteration (NeurIPS 2023 · similarity 0.89)
- Efficient Communication in Multi-Agent Reinforcement Learning with Implicit Consensus Generation (AAAI 2025 · similarity 0.88)
- Cautiously-Optimistic Knowledge Sharing for Cooperative Multi-Agent Reinforcement Learning (AAAI 2024 · similarity 0.88)
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.