Timezone: »

 
Spotlight
Langevin Monte Carlo for Contextual Bandits
Pan Xu · Hongkai Zheng · Eric Mazumdar · Kamyar Azizzadenesheli · Animashree Anandkumar

Tue Jul 19 08:30 AM -- 08:35 AM (PDT) @ Hall F
in Reinforcement Learning »

We study the efficiency of Thompson sampling for contextual bandits. Existing Thompson sampling-based algorithms need to construct a Laplace approximation (i.e., a Gaussian distribution) of the posterior distribution, which is inefficient to sample in high dimensional applications for general covariance matrices. Moreover, the Gaussian approximation may not be a good surrogate for the posterior distribution for general reward generating functions. We propose an efficient posterior sampling algorithm, viz., Langevin Monte Carlo Thompson Sampling (LMC-TS), that uses Markov Chain Monte Carlo (MCMC) methods to directly sample from the posterior distribution in contextual bandits. Our method is computationally efficient since it only needs to perform noisy gradient descent updates without constructing the Laplace approximation of the posterior distribution. We prove that the proposed algorithm achieves the same sublinear regret bound as the best Thompson sampling algorithms for a special case of contextual bandits, viz., linear contextual bandits. We conduct experiments on both synthetic data and real-world datasets on different contextual bandit models, which demonstrates that directly sampling from the posterior is both computationally efficient and competitive in performance.

Keywords: Reinforcement Learning · MISC: Online Learning, Active Learning and Bandits

Author Information

Pan Xu (California Institute of Technology)
Hongkai Zheng (Caltech)
Eric Mazumdar (Caltech)
Kamyar Azizzadenesheli (Purdue University)
Animashree Anandkumar (Caltech and NVIDIA)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors