Reader, Tata Institute of Fundamental Research
2 papers at NeurIPS 2025
An algorithm with a static regret of $O(\sqrt{T})$ and a CCV of $\min\{{\cal V}, O(\sqrt{T}\log T) \}$, for constrained online convex optimization where ${\cal V}$ depends on the geometric properties of the instance .
This paper presents a tunable algorithm for online convex optimization with adversarial constraints that significantly reduces cumulative constraint violation below $O(\sqrt{T})$ by trading it off with regret.
