Allocation Stability and Wald Inference under Variance-Aware UCB

Name: Allocation Stability and Wald Inference under Variance-Aware UCB
Uploaded: 2026-10-08T20:39:51.953390+00:00
Duration: 807 s
Description: This work precisely characterizes arm-pulling rates for Variance-Aware UCB (UCB-V), showing its performance depends critically on variance ratios and exploration coefficients. While UCB-V achieves a refined regret bound, it can become unstable under specific conditions, highlighting the need for new statistical inference methods to ensure reliable decision-making in adaptive sampling.

This work precisely characterizes arm-pulling rates for Variance-Aware UCB (UCB-V), showing its performance depends critically on variance ratios and exploration coefficients. While UCB-V achieves a refined regret bound, it can become unstable under specific conditions, highlighting the need for new statistical inference methods to ensure reliable decision-making in adaptive sampling.

Open the full episode