2018 (modified: 19 May 2025)ICML 2018Readers: Everyone
Abstract:Thompson sampling has impressive empirical performance for many multi-armed bandit problems. But current algorithms for Thompson sampling only work for the case of conjugate priors since they requi...