NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:609
Title:Polynomial Cost of Adaptation for X-Armed Bandits

The reviewers are in agreement that the paper makes good progress on adaptivity of x-armed bandit problems, specifically providing an agnostic algorithm that achieves Pareto optimal regret rates, which matches regret lower bounds for adaptive algorithms. The algorithm that achieves this goal is also of independent interest.