Research map / Multi-agent systems and games / Games and equilibria
Finding Friend and Foe in Multi-Agent Games
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 |
|---|---|---|---|
| add DeepRole instead of a human player | a team of 4 DeepRole agents · ProAvalon.com, added to 4 DeepRole | overall win rate: 38.8 ± 1.3 -> 46.9 ± 0.6 | Helped |
| add DeepRole instead of a human player | a team of 4 DeepRole agents · ProAvalon.com, added to 4 DeepRole | Resistance win rate: 25.6 ± 1.5 -> 34.4 ± 0.7 | Helped |
| add DeepRole instead of a human player | a team of 4 DeepRole agents · ProAvalon.com, added to 4 DeepRole | Spy win rate: 57.8 ± 2.0 -> 65.6 ± 0.9 | Helped |
| add DeepRole instead of a human player | a team of 4 humans · ProAvalon.com, added to 4 Human | overall win rate: 48.1 ± 1.2 -> 60.0 ± 5.5 | Helped |
| add DeepRole instead of a human player | a team of 4 humans · ProAvalon.com, added to 4 Human | Resistance win rate: 40.3 ± 1.5 -> 51.4 ± 8.2 | Helped |
| add DeepRole instead of a human player | a team of 4 humans · ProAvalon.com, added to 4 Human | Spy win rate: 59.7 ± 1.9 -> 67.4 ± 7.1 | Helped |
| replace the win probability layer with a zero-sum layer (ablation) | DeepRole value networks · Value-network held-out samples | generalization loss | Hurt |
| replace the win probability layer with a zero-sum layer (ablation) | DeepRole · Five-player Avalon | agent strength | Hurt |
| remove logical deduction and use a zero-sum value layer (ablation) | DeepRole · Five-player Avalon | agent strength | Hurt |
| increase CFR solving iterations per move (ablation) | DeepRole · Five-player Avalon | meta-game basin of attraction | Helped |
Closest papers by meaning
- Towards Playing Full MOBA Games with Deep Reinforcement Learning (NeurIPS 2020 · similarity 0.87)
- Mimicking To Dominate: Imitation Learning Strategies for Success in Multiagent Games (NeurIPS 2024 · similarity 0.86)
- Self-Play $Q$-Learners Can Provably Collude in the Iterated Prisoner's Dilemma (ICML 2025 · similarity 0.85)
- Outbidding and Outbluffing Elite Humans: Mastering Liar’s Poker via Self-Play and Reinforcement Learning (NeurIPS 2026 · similarity 0.85)
- A Generalist Hanabi Agent (ICLR 2025 · similarity 0.85)
- MEMO: Memory-Augmented Model Context Optimization for Robust Multi-Turn Multi-Agent LLM Games (ICML 2026 · 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.