Timezone: »

Poster
Instance Dependent Regret Analysis of Kernelized Bandits
Shubhanshu Shekhar · Tara Javidi

Thu Jul 21 03:00 PM -- 05:00 PM (PDT) @ Hall E #720
We study the kernelized bandit problem, that involves designing an adaptive strategy for querying a noisy zeroth-order-oracle to efficiently learn about the optimizer of an unknown function $f$ with a norm bounded by $M<\infty$ in a Reproducing Kernel Hilbert Space~(RKHS) associated with a positive definite kernel $K$. Prior results, working in a \emph{minimax framework}, have characterized the worst-case~(over all functions in the problem class) limits on regret achievable by \emph{any} algorithm, and have constructed algorithms with matching~(modulo polylogarithmic factors) worst-case performance for the Matern family of kernels. These results suffer from two drawbacks. First, the minimax lower bound gives no information about the limits of regret achievable by commonly used algorithms on specific problem instances. Second, due to their worst-case nature, the existing upper bound analysis fails to adapt to easier problem instances within the function class. Our work takes steps to address both these issues. First, we derive \emph{instance-dependent} regret lower bounds for algorithms with uniformly~(over the function class) vanishing normalized cumulative regret. Our result, valid for all the practically relevant kernelized bandits algorithms, such as, GP-UCB, GP-TS and SupKernelUCB, identifies a fundamental complexity measure associated with every problem instance. We then address the second issue, by proposing a new minimax near-optimal algorithm that also adapts to easier problem instances.

#### More from the Same Authors

• 2020 Poster: Adaptive Sampling for Estimating Probability Distributions »