Multi-Player Approaches for Dueling Bandits

2024-05-25

arXiv.org

AI Summary

This work addresses the challenge of collaborative exploration of non-informative arm pairs in multi-player preference-feedback dueling bandits. Methodologically, it introduces the first theoretically optimal distributed solution: (i) it adapts the Follow-Your-Leader black-box framework to the multi-player dueling setting, achieving the fundamental regret lower bound; and (ii) it designs a distributed protocol leveraging message passing and Condorcet winner recommendation to enable efficient coordination. The approach integrates techniques from multi-agent reinforcement learning, Double Thompson Sampling, and distributed decision-making. Empirical evaluation across multiple benchmark tasks demonstrates that the proposed algorithm significantly outperforms single-player baselines—achieving 37%–62% faster convergence and reducing cumulative regret by 41%–58%. These results validate its dual advantage: rigorous theoretical guarantees and superior practical performance.

Technology Category

• Machine Learning: Online Learning & Bandits
• Multiagent Systems: Adversarial Agents
• Game Theory and Economic Paradigms: Adversarial Learning

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Fairness, privacy, and diversity in economic environments
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Various approaches have emerged for multi-armed bandits in distributed systems. The multiplayer dueling bandit problem, common in scenarios with only preference-based information like human feedback, introduces challenges related to controlling collaborative exploration of non-informative arm pairs, but has received little attention. To fill this gap, we demonstrate that the direct use of a Follow Your Leader black-box approach matches the lower bound for this setting when utilizing known dueling bandit algorithms as a foundation. Additionally, we analyze a message-passing fully distributed approach with a novel Condorcet-winner recommendation protocol, resulting in expedited exploration in many cases. Our experimental comparisons reveal that our multiplayer algorithms surpass single-player benchmark algorithms, underscoring their efficacy in addressing the nuanced challenges of the multiplayer dueling bandit setting.

Problem

• Addressing multiplayer dueling bandits with preference-based feedback
• Optimizing collaborative exploration of non-informative arm pairs
• Enhancing distributed algorithms for Condorcet-winner recommendation protocols

Innovation

• Follow Your Leader black-box approach
• message-passing fully distributed approach
• Condorcet-winner recommendation protocol