Details
- Time: 10:00 -11:00 am (Beijing Time)
- Date: Friday, March 22, 2024
- Venue: MB441
- Speaker: Prof. Min Gan (Qingdao University)
Abstract
Many tasks in machine learning can be deduced to an optimization problem, which are often structured, that is, some parameters are relatively easy to optimize. The separable nonlinear least squares problem is a typical instance of these kinds of problems. This talk first provides an efficient algorithm for optimizing separable nonlinear least squares problems --- variable projection algorithm, then introduces the generalized separable problems and their solutions, and finally provides a stochastic approximation (learning) algorithm for separable stochastic optimization problems.
Speaker
Min Gan, PhD, Professor and doctoral supervisor in the School of Computer Science and Technology at Qingdao University and a senior member of IEEE. He obtained his PhD degree in Control Science and Engineering from Central South University in 2010. His primary research interests lie in machine learning and computer vision. He has published over 30 articles in key journals in the field of computers and automation.