description [ICLR 2026][Learning Theory][Single-index model] This paper proves that when learning single-index models in an interactive (bandit) environment, a naive normalized SGD equipped with ...

Highlights & Insights Theory-Practice Unification: BLAE is the first batched linear bandit algorithm that achieves both minimax optimal regret (in both regimes) and outstanding practical performance, ...

Abstract: This letter provides the first rigorous theoretical analysis for the Distributed Fenchel Dual Gradient (DFDG) algorithm, a continuous-time method for solving distributed convex optimization ...

Abstract: The parallel efficient global optimization (EGO) algorithm was developed to leverage the rapid advancements in high-performance computing. However, conventional parallel EGO algorithm based ...