Stochastic optimization problems with deterministic constraints commonly appear in machine learning, finance, and engineering applications. This talk presents a new adaptive solution strategy for this important class of problems. The aim is to decrease the computational cost while maintaining an optimal convergence rate. The guiding principle is to adjust the batch size (or sample size) on the fly so that the error in the gradient approximation remains proportional to the error in the underlying optimization problem. After providing motivation and context, I will present the first adaptive sampling algorithms for constrained optimization under uncertainty that simultaneously maintain optimal sample efficiency and iteration complexity. I will then demonstrate the efficacy of this new adaptive sampling strategy in multiple applications, drawing mainly from use cases found in engineering design. This talk aims to be accessible and inviting to a broad mathematical audience.

Zoom Meeting: Link
Meeting ID: 976 3868 1103
Passcode: NAseminar