-
Max-Quantile Grouped Infinite-Arm Bandits
In recommendation systems, one may wish to know which among several populations has the highest median click-through rate, while displaying as few total recommendations as possible. In this paper, we develop an algorithm to identify the infinite-arm group whose reservoir distribution has the highest median.
-
Analysis of Sampling for Diffusion Models
In this project, we conduct an empirical analysis of sampling algorithms, focusing on Langevin algorithms, integral to generative models like Diffusion models. Our primary objective is to compare the runtimes of different sampling methods and explore enhancements for more efficient sampling procedures.