Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about optimal algorithms for online convex optimization with adversarial constraints in this seminar talk by Prof. Abhishek Sinha from TIFR, Mumbai. The lecture addresses a long-standing open question in constrained online convex optimization (COCO) by demonstrating that an online policy can simultaneously achieve O(√T) regret and Õ(√T) cumulative constraint violation against an adaptive adversary. Discover how combining the adaptive regret bound of the AdaGrad algorithm with Lyapunov optimization from control theory leads to these breakthrough results, including improved O(log T) regret guarantees for strongly convex cost functions. Prof. Sinha, recipient of multiple prestigious awards including the Google India Research Award 2023 and INSA Medal for Young Scientists 2021, presents this elegant solution in the field of theoretical machine learning. The seminar is scheduled for May 21, 2025, and is organized by the Centre for Networked Intelligence at IISc.
