# Generating and Adapting to Diverse Ad-Hoc Cooperation Agents in Hanabi

arXiv: https://arxiv.org/abs/2004.13710 · HTML: https://mlsanity.com/paper/2004.13710/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 28 Apr 2020.

## Built on by (most cited first, 13 of 13)

- [JaxMARL: Multi-Agent RL Environments and Algorithms in JAX](https://mlsanity.com/paper/2311.10090/) (2023)
- [Cooperative Open-ended Learning Framework for Zero-shot Coordination](https://mlsanity.com/paper/2302.04831/) (2023)
- [The Collaboration Gap: Exploration and Benchmarking of Open-World Agentic Cooperation](https://mlsanity.com/paper/2511.02687/) (2025)
- [Tackling Cooperative Incompatibility for Zero-Shot Human-AI Coordination](https://mlsanity.com/paper/2306.03034/) (2023)
- [A Minimax Approach to Ad Hoc Teamwork](https://mlsanity.com/paper/2502.02377/) (2025)
- [Symmetry-Breaking Augmentations for Ad Hoc Teamwork](https://mlsanity.com/paper/2402.09984/) (2024)
- [Adapting to Teammates in a Cooperative Language Game](https://mlsanity.com/paper/2403.00823/) (2024)
- [HOLA-Drone: Hypergraphic Open-ended Learning for Zero-Shot Multi-Drone Cooperative Pursuit](https://mlsanity.com/paper/2409.08767/) (2024)
- [Two-Step Reinforcement Learning for Multistage Strategy Card Game](https://mlsanity.com/paper/2311.17305/) (2023)
- [AT-Drone: Benchmarking Adaptive Teaming in Multi-Drone Pursuit](https://mlsanity.com/paper/2502.09762/) (2025)
- [Enhancing Player Enjoyment with a Two-Tier DRL and LLM-Based Agent System for Fighting Games](https://mlsanity.com/paper/2504.07425/) (2025)
- [NestRL: A Nested Training Regime for Mutual Adaptation in Human-AI Teaming](https://mlsanity.com/paper/2602.17737/) (2026)
- [Smart Alone, Blind Together: Why Accuracy Fails to CertifyTheory of Mind in Multi-Agent Teams](https://mlsanity.com/paper/2610.19051/) (2026)

## Abstract (arXiv)

Hanabi is a cooperative game that brings the problem of modeling other players to the forefront. In this game, coordinated groups of players can leverage pre-established conventions to great effect, but playing in an ad-hoc setting requires agents to adapt to its partner's strategies with no previous coordination. Evaluating an agent in this setting requires a diverse population of potential partners, but so far, the behavioral diversity of agents has not been considered in a systematic way. This paper proposes Quality Diversity algorithms as a promising class of algorithms to generate diverse populations for this purpose, and generates a population of diverse Hanabi agents using MAP-Elites. We also postulate that agents can benefit from a diverse population during training and implement a simple "meta-strategy" for adapting to an agent's perceived behavioral niche. We show this meta-strategy can work better than generalist strategies even outside the population it was trained with if its partner's behavioral niche can be correctly inferred, but in practice a partner's behavior depends and interferes with the meta-agent's own behavior, suggesting an avenue for future research in characterizing another agent's behavior during gameplay.