Abstract: We consider a Latent Bandit problem where the latent state keeps changing in time according to an underlying Markov chain, and every state is represented by a specific Bandit instance. At each step, the agent chooses an arm and observes a random reward but is unaware of which MAB he is currently pulling. As typical in Latent Bandits, we assume to know the reward distribution of the arms of all the Bandit instances. Within this setting, our goal is to learn the transition matrix determined by the Markov process.
We propose a technique to tackle this estimation problem that results in solving a least-square problem obtained by exploiting the knowledge of the reward distributions and the properties of Markov chains. We prove the consistency of the estimation procedure, and we make a theoretical comparison with standard Spectral Decomposition techniques. We then discuss the dependency of the problem on the number of arms and present an offline method that chooses the best subset of possible arms that can be used for the estimation of the transition model. We ultimately introduce the SL-EC algorithm based on an Explore then Commit strategy that uses the proposed approach to estimate the transition model during the exploration phase. This algorithm achieves a regret of the order $\widetilde{\mathcal{O}}(T^{2/3})$ when compared against an oracle that builds a belief representation of the current state using the knowledge of both the observation and transition model and optimizes the expected instantaneous reward at each step. Finally, we illustrate the effectiveness of the approach and compare it with state-of-the-art algorithms for non-stationary bandits and with a modified technique based on spectral decomposition.
Submission Length: Long submission (more than 12 pages of main content)
Previous TMLR Submission Url: https://openreview.net/forum?id=qkKgjX5Rr1
Changes Since Last Submission: The main changes with respect to the previous version have been to rewrite the formulation of the problem and the sections describing the estimation procedure. We adopted a new notation in order to be more clear and remove the previous ambiguities.
We revised the structure of the work. Among the variations, we added a new section in the Appendix giving more details on the optimization of the exploration length $T_0$ when the information about some parameters is not available; we moved the theoretical section comparing our approach with spectral decomposition methods from the main paper to the Appendix.
Code: https://github.com/alesnow97/SwitchingLatentBandits
Supplementary Material: pdf
Assigned Action Editor: ~Gergely_Neu1
Submission Number: 2315
Loading