AI RESEARCH

Stochastic Regret Guarantees for Online Zeroth- and First-Order Bilevel Optimization

arXiv CS.LG

ArXi:2511.01126v2 Announce Type: replace Online bilevel optimization (OBO) is a powerful framework for machine learning problems where both outer and inner objectives evolve over time, requiring dynamic updates. Current OBO approaches rely on deterministic \textit{window-smoothed} regret minimization, which may not accurately reflect system performance when functions change rapidly. In this work, we