一种求解稀疏逻辑回归问题的不精确邻近拟牛顿算法  

An inexact proximal quasi-Newton algorithm for solving sparse logistic regression problems

作　　者：杨娅 肖斌 胡清洁[1] YANG Ya;XIAO Bin;HU Qingjie(School of Mathematics and Computing Science,Guilin University of Electronic Technology,Guilin 541004,China;General Education Center,Zhengzhou Business University,Zhengzhou 451200,China)

机构地区：[1]桂林电子科技大学数学与计算科学学院,广西桂林541004 [2]郑州商学院通识教育中心,郑州451200

出　　处：《桂林电子科技大学学报》2021年第3期241-246,共6页Journal of Guilin University of Electronic Technology

基　　金：国家自然科学基金(11961011)。

摘　　要：复合凸优化问题是一类重要的优化问题,它在图像去模糊、人脸识别和压缩感知等领域具有广泛的应用。针对稀疏逻辑回归问题,提出了一种不精确加速邻近拟牛顿算法,并给出了该算法的收敛速度分析。数值结果表明,不精确加速邻近拟牛顿算法求解稀疏逻辑回归问题是有效的。Convex composite optimization problem is an important optimization problems,which is widely used in image deblurring,face recognition and compressed sensing and so on.Aiming at the sparse logic regression problem,an inexact accelerated proximal quasi-newton algorithm for sparse logic regression problem is proposed,and the convergence rate analysis of the algorithm is given.The numerical results show that the inexact accelerated proximal quasi-Newton algorithm is effective in solving sparse logic regression problems.

关 键 词：稀疏逻辑回归问题 加速邻近拟牛顿算法 不精确邻近拟牛顿算法 

分 类 号：O224[理学—运筹学与控制论]