An Alternating Proximal Splitting Method with Global Convergence for Nonconvex Structured Sparsity Optimization

Authors

  • Shubao Zhang Zhejiang University
  • Hui Qian Zhejiang University
  • Xiaojin Gong Zhejiang University

DOI:

https://doi.org/10.1609/aaai.v30i1.10253

Abstract

In many learning tasks with structural properties, structured sparse modeling usually leads to better interpretability and higher generalization performance. While great efforts have focused on the convex regularization, recent studies show that nonconvex regularizers can outperform their convex counterparts in many situations. However, the resulting nonconvex optimization problems are still challenging, especially for the structured sparsity-inducing regularizers. In this paper, we propose a splitting method for solving nonconvex structured sparsity optimization problems. The proposed method alternates between a gradient step and an easily solvable proximal step, and thus enjoys low per-iteration computational complexity. We prove that the whole sequence generated by the proposed method converges to a critical point with at least sublinear convergence rate, relying on the Kurdyka-Łojasiewicz inequality. Experiments on both simulated and real-world data sets demonstrate the efficiency and efficacy of the proposed method.

Downloads

Published

2016-03-02

How to Cite

Zhang, S., Qian, H., & Gong, X. (2016). An Alternating Proximal Splitting Method with Global Convergence for Nonconvex Structured Sparsity Optimization. Proceedings of the AAAI Conference on Artificial Intelligence, 30(1). https://doi.org/10.1609/aaai.v30i1.10253

Issue

Section

Technical Papers: Machine Learning Methods