Plenary Speaker 1

Reinforcement Learning and Adaptive Subspace Assisted Zeroth-Order Optimization Solvers for Improved Robustness

Zeroth-order optimization solvers are often deployed in settings where little information regarding a problem’s conditioning or noise level is known. An ideal solver will perform well in a variety of challenging settings. We report on our experience developing adaptive algorithms that leverage information learned online to adapt critical algorithmic features. We illustrate our approach in trust-region-based reduced-space methods, which significantly improve scalability for large dimensional problems. We show how trained policies can even be deployed effectively in nonstationary cases, where the noise seen changes across the decision space.

This talk is joint work with Pengcheng Xie and Kwassi Joseph Dzahini.