Sparsity Optimization and Regularization Method

Dr. Kazufumi Ito

Department of Mathematics
North Carolina State University

18 March 2011
218 Shelby Center
3:00 (Refreshements at 2:30)

Abstract

Many of applications including structural design, inverse problem, state estimation, classification, deconvolution and inverse scattering problems, bionetworks, signal-image analysis and compression can be formulated as the regularized minimization problem. The regularization methodology based on the sparsity measure (numbers of nonzeros) is discussed and analyzed for such problems. A general mathematical formulation as well as numerical algorithms to solve a general class of sparsity optimization are discussed. Some numerical examples are also presented to demonstrate the capability and applicability of our approach.