Google Research

Randomized Exploration in Generalized Linear Bandits

NeurIPS 2019 (2020) (to appear)

Abstract

We study two randomized algorithms for generalized linear bandits, GLM-TSL and GLM-FPL. GLM-TSL samples a generalized linear model (GLM) from the Laplace approximation to the posterior distribution. GLM-FPL, a new algorithm proposed in this work, fits a GLM to a randomly perturbed history of past rewards. We prove a $\tilde{O}(d \sqrt{n} + d^2)$ upper bound on the $n$-round regret of GLM-TSL, where $d$ is the number of features. This is the first regret bound of a Thompson sampling-like algorithm in GLM bandits where the leading term is $\tilde{O}(d \sqrt{n})$. We apply both GLM-TSL and GLM-FPL to logistic and neural network bandits, and show that they perform well empirically. In more complex models, GLM-FPL is significantly faster. Our results showcase the role of randomization, beyond posterior sampling, in exploration.

Research Areas

Learn more about how we do research

We maintain a portfolio of research projects, providing individuals and teams the freedom to emphasize specific types of work