[The 78th TrustML Young Scientist Seminar] Talk by Dheeraj Baby (UC Santa Barbara)

This is an online seminar.
Registration is required.

Title:
Online Label Shift: Optimal Dynamic Regret meets Practical Algorithms

Speaker:
Dheeraj Baby (UC Santa Barbara)

Abstract:
This talk will focus on the unsupervised online label shift problem where the class marginals vary but the class-conditionals remain invariant. Our goal is to adapt a learner, trained on some offline labeled data, to changing label distributions given unlabeled online data. We develop novel algorithms that reduce the adaptation problem to online regression and guarantee optimal dynamic regret without any prior knowledge of the extent of drift in the label distribution. Our solution is based on bootstrapping the estimates of online regression oracles that track the drifting proportions. We will also present experimental results that back up the efficacy of the proposed method.

Bio:
Dheeraj is a final year PhD student at UC Santa Barbara. His research has been focused on developing theory and algorithms for dynamic regret minimization under losses with curved geometry. Applications of his work include time-series forecasting, handling distribution shift, non-parametric regression and dynamic pricing. He has earned accolades, including a Best Student Paper award at COLT'21 and a spotlight at NeurIPS'23.