English

Iterative thresholding algorithm based on non-convex method for modified lp-norm regularization minimization

Optimization and Control 2018-04-26 v1

Abstract

Recently, the \lp\l_{p}-norm regularization minimization problem (Ppλ)(P_{p}^{\lambda}) has attracted great attention in compressed sensing. However, the \lp\l_{p}-norm xpp\|x\|_{p}^{p} in problem (Ppλ)(P_{p}^{\lambda}) is nonconvex and non-Lipschitz for all p(0,1)p\in(0,1), and there are not many optimization theories and methods are proposed to solve this problem. In fact, it is NP-hard for all p(0,1)p\in(0,1) and λ>0\lambda>0. In this paper, we study two modified \lp\l_{p} regularization minimization problems to approximate the NP-hard problem (Ppλ)(P_{p}^{\lambda}). Inspired by the good performance of Half algorithm and 2/32/3 algorithm in some sparse signal recovery problems, two iterative thresholding algorithms are proposed to solve the problems (Pp,1/2,ϵλ)(P_{p,1/2,\epsilon}^{\lambda}) and (Pp,2/3,ϵλ)(P_{p,2/3,\epsilon}^{\lambda}) respectively. Numerical results show that our algorithms perform effectively in finding the sparse signal in some sparse signal recovery problems for some proper p(0,1)p\in(0,1).

Keywords

Cite

@article{arxiv.1804.09385,
  title  = {Iterative thresholding algorithm based on non-convex method for modified lp-norm regularization minimization},
  author = {Angang Cui and Jigen Peng and Haiyang Li and Meng Wen and Jiajun Xiong},
  journal= {arXiv preprint arXiv:1804.09385},
  year   = {2018}
}