A unified model for the sparse optimal scoring problem

作者：

Highlights：

• We derive an efficient iterative algorithm based on alternating direction method of multipliers (ADMM) for the new formulation. The new proposed method can solve (Lo−) norm regularized problem directly rather than using convex or nonconvex approximations of (Lo−) norm.

• The convergence of the algorithm is analyzed theoretically.

• Extensive numerical experiments show that our algorithm is efficient not only in classification accuracy but also in sparsity.

摘要

•A unified model for sparse optimal scoring is proposed by employing Lu-norm (0≤q≤1) regular term where Lo−norm and(Lu) -norm (0≤q≤1) will be selected adaptively to find more sparser solutions.•We derive an efficient iterative algorithm based on alternating direction method of multipliers (ADMM) for the new formulation. The new proposed method can solve (Lo−) norm regularized problem directly rather than using convex or nonconvex approximations of (Lo−) norm.•The convergence of the algorithm is analyzed theoretically.•Extensive numerical experiments show that our algorithm is efficient not only in classification accuracy but also in sparsity.

论文关键词：Optimal scoring,Linear discriminant analysis,Feature selection,ℓq−norm,Sparseness

论文评审过程：Received 14 February 2022, Revised 4 August 2022, Accepted 13 August 2022, Available online 18 August 2022, Version of Record 28 August 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108976