Yu-Xiang Wang & Dheeraj Baby

By Caroline Mecartea, Communications & Special Events 

A paper titled “Optimal Dynamic Regret in Exp-Concave Online Learning” written by Assistant Professor Yu-Xiang Wang and PhD candidate Dheeraj Baby was selected as one of two best student papers at COLT 2021. The Conference on Learning Theory is widely regarded as the top conference for machine learning theory. The paper is to be presented LIVE in a plenary session on Monday August 16.

The paper that receives the award — "Optimal Dynamic Regret in Exp-Concave Online Learning” — addresses a 17-year-old open problem due to (Zinkevich, 2003) and the developed techniques have profound impacts on the theory of machine learning in non-stationary and adversarial environments as well as in nonparametric statistics and time series forecasting. 

Congratulations Yu-Xiang Wang & Dheeraj Baby!