Thời gian: 9h, Sáng thứ Ba, ngày 14/12/2021

Link tham dự seminar: https://meet.google.com/gtq-datz-zwa?pli=1&authuser=1

Tóm tắt: In these talks, we discuss about proximal point method, a ``conceptual” algorithm for minimizing a lower semicontinuous and convex function, which is the basis of various methods, such as proximal gradient method, Levenberg-Marquardt algorithm, augmented Lagrangian method, alternating direction method of multipliers, and primal-dual hybrid gradient method for solving convex composite optimization problems. We will have a closer look at these methods‘ interpretation and convergence when being applied to convex composite optimization problems typically arising in machine learning and data science areas.