36-772 Martingales 2: Concentration Inequalities, Applications to Sequential Analysis

6 units, Carnegie Mellon University

This second mini will focus on deriving guarantees for a variety of important problems in sequential analysis using the tools developed in the first mini, as well as new tools such as uniform nonasymptotic versions of the law of the iterated logarithm for scalars and matrices. Applications include sequential analogs of the ttest, that are valid without a Gaussian assumption, bestarm identification in multiarmed bandits, average treatment effect estimation in sequential clinical trials, sequential covariance matrix estimation, and other such problems.

Prerequisites: none listed

Courseweave needs JavaScript to show sections, workloads and planning tools.