Numerical optimization algorithms for large-scale problems in machine learning and control

The topic of this PhD is to develop novel numerical optimization algorithms for solving large-scale problems arising in machine learning and control. The focus of the thesis will be threefold: 1. Theory: global convergence analysis, asymptotic rate analysis, saddle point avoidance for nonconvex problems, 2. Implementation: efficient numerical linear algebra routines, open-source software, 3. Applications: deep restricted kernel machines, distributionally robust machine learning and control.