WFVML 2022 Invited Talk: Computational Methods for Non-convex Machine Learning Problems

Prof. Somayeh Sojoudi (UC Berkeley)

Somayeh Sojoudi is an Assistant Professor in the Departments of Electrical Engineering & Computer Sciences and Mechanical Engineering at the University of California, Berkeley. She is an Associate Editor for the journals of the IEEE Transactions on Smart Grid, Systems & Control Letters, IEEE Access, and IEEE Open Journal of Control Systems. She is also a member of the conference editorial board of the IEEE Control Systems Society. She received several awards and honors, including INFORMS Optimization Society Prize for Young Researchers, INFORMS Energy Best Publication Award, INFORMS Data Mining Best Paper Award, NSF CAREER Award, and ONR Young Investigator Award. She has also received several best student conference paper awards (as advisor or co-author) from the Control Systems Society.

Invited Talk: Computational Methods for Non-convex Machine Learning Problems

We need efficient computational methods with provable guarantees that can cope with the complex nature and high nonlinearity of many real-world systems. Practitioners often design heuristic algorithms tailored to specific applications, but the theoretical underpinnings of these methods remain a mystery, and this limits their usage in safety-critical systems. In this talk, we aim to address the above issue for some machine learning problems. First, we study the problem of certifying the robustness of neural networks against adversarial inputs. Then we study when simple local search algorithms could solve a class of nonlinear problems to global optimality. We discuss our recent results in addressing these problems and demonstrate them on tensor decomposition with outliers and video processing.