EE Systems Seminar
Weakly convex optimization for signal processing problems
Abstract Motivated by the fact that a weakly convex function as an approximation of the l0-norm is able to better induce sparsity than that of l1-norm, several weakly convex optimization algorithms based on the Alternating Direction Method of Multipliers (ADMM) have been proposed for various signal processing problems, including sparse signal recovery and direction-of-arrival estimation etc.. However, the effectiveness of such algorithm is severely weakened when the signal is being corrupted by non-Gaussian noise that generates observation outliers. We proposed to apply proximal operators to the weakly convex functions to suppress the outliers. Simulation results showed that the proposed algorithm has high outlier resistance and can enjoy much better performance.
Bio Qi Liu received the B.E. degree in Measuring & Control Technology and Instrumentation and M.Sc. degree in Control Science and Engineering from College of Automation, Harbin Engineering University, Harbin, China, in 2013 and 2016, respectively. Currently, he is studying for Ph.D. degree in the Department of Electronic Engineering, City University of Hong Kong, under the guidance of Prof. H.C. So and Dr. Chi-Wah Kok. His research interests include robust compressed sensing and its applications to sparse signal recovery, parameter estimation and image denoising/inpainting.
Contact: Liliana Chavarria at 626-395-4715 email@example.com