Abstract

 In this paper, we propose a Double Thompson Sampling (D-TS) algorithm for dueling bandit problems. As its name suggests, D-TS selects both the first and the second candidates according to Thompson Sampling. Specifically, D-TS maintains a posterior distribution for the preference matrix, and chooses the pair of arms for comparison according to two sets of samples independently drawn from the posterior distribution. This simple algorithm applies to general Copeland dueling bandits, including Condorcet dueling bandits as a special case. For general Copeland dueling bandits, we show that D-TS achieves  regret. Moreover, using a back substitution argument, we refine the regret to  in Condorcet dueling bandits and most practical Copeland dueling bandits. In addition, we propose an enhancement of D-TS, referred to as D-TS+ , to reduce the regret in practice by carefully breaking ties. Experiments based on both synthetic and real-world data demonstrate that D-TS and D-TS+ significantly improve the overall performance, in terms of regret and robustness.