Nonconvex Regularization Techniques for Sparse Signal Recovery

Event Start
Event End


To recover a sparse signal from the linear system Ax = b, a convex lp regularization method (i.e., 1 ≤ p < 2) is commonly used under certain conditions. Recently, however, more attentions have been paid to use the nonconvex lp regularization method (i.e., 0 < p < 1, in particular, p = 1/2) to recover a sparse signal. In this talk, we will discuss a nonconvex reweighted lp regularization method for recovering a sparse signal from the linear system. Our model is the nonconvex version of the model of S. Voronin and I. Daubechies (An iteratively reweighted least squares algorithm for sparse regularization, arXiv:1511.08970v3).

Brief Biography

Hong-Kun Xu, currently Distinguished Professor at Hangzhou Dianzi University, is a member of the Academy of Science of South Africa and Fellow to The World Academy of Sciences (TWAS). He was selected highly cited researcher (Thomson Reuters, 2014-2016 and Clarivate Analytics, 2017-2018). His research interest includes nonlinear functional analysis, fixed point theory and optimization.