Convergence guarantees for a class of non-convex and non-smooth optimization problems

被引:0
|
作者
Khamaru, Koulik [1 ]
Wainwright, Martin J. [1 ,2 ,3 ]
机构
[1] Univ Calif Berkeley, Dept Stat, Berkeley, CA 94720 USA
[2] Univ Calif Berkeley, Dept Elect Engn & Comp Sci, Berkeley, CA 94720 USA
[3] Voleon Grp, Berkeley, CA USA
基金
美国国家科学基金会;
关键词
ALGORITHM; MINIMIZATION; CONVEXITY;
D O I
暂无
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
We consider the problem of finding critical points of functions that are non-convex and non-smooth. Studying a fairly broad class of such problems, we analyze the behavior of three gradient-based methods (gradient descent, proximal update, and Frank-Wolfe update). For each of these methods, we establish rates of convergence for general problems, and also prove faster rates for continuous sub-analytic functions. We also show that our algorithms can escape strict saddle points for a class of non-smooth functions, thereby generalizing known results for smooth functions. Our analysis leads to a simplification of the popular CCCP algorithm, used for optimizing functions that can be written as a difference of two convex functions. Our simplified algorithm retains all the convergence properties of CCCP, along with a significantly lower cost per iteration. We illustrate our methods and theory via applications to the problems of best subset selection, robust estimation, mixture density estimation, and shape-from-shading reconstruction.
引用
收藏
页数:52
相关论文
共 50 条
  • [1] Convergence guarantees for a class of non-convex and non-smooth optimization problems
    Khamaru, Koulik
    Wainwright, Martin J.
    Journal of Machine Learning Research, 2019, 20
  • [2] Convergence guarantees for a class of non-convex and non-smooth optimization problems
    Khamaru, Koulik
    Wainwright, Martin J.
    INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 80, 2018, 80
  • [3] Learning an Alternating Bergman Network for Non-convex and Non-smooth Optimization Problems
    Wang, Yiyang
    Liu, Risheng
    Su, Zhixun
    INTELLIGENCE SCIENCE AND BIG DATA ENGINEERING, ISCIDE 2017, 2017, 10559 : 11 - 27
  • [4] Nested Alternating Minimization with FISTA for Non-convex and Non-smooth Optimization Problems
    Gur, Eyal
    Sabach, Shoham
    Shtern, Shimrit
    JOURNAL OF OPTIMIZATION THEORY AND APPLICATIONS, 2023, 199 (03) : 1130 - 1157
  • [5] Nested Alternating Minimization with FISTA for Non-convex and Non-smooth Optimization Problems
    Eyal Gur
    Shoham Sabach
    Shimrit Shtern
    Journal of Optimization Theory and Applications, 2023, 199 : 1130 - 1157
  • [6] Efficient Convex Optimization for Non-convex Non-smooth Image Restoration
    Li, Xinyi
    Yuan, Jing
    Tai, Xue-Cheng
    Liu, Sanyang
    JOURNAL OF SCIENTIFIC COMPUTING, 2024, 99 (02)
  • [7] ON A NEW SMOOTHING TECHNIQUE FOR NON-SMOOTH, NON-CONVEX OPTIMIZATION
    Yilmaz, Nurullah
    Sahiner, Ahmet
    NUMERICAL ALGEBRA CONTROL AND OPTIMIZATION, 2020, 10 (03): : 317 - 330
  • [8] KKT OPTIMALITY CONDITIONS IN NON-SMOOTH, NON-CONVEX OPTIMIZATION
    Sisarat, Nithirat
    Wangkeeree, Rabian
    Lee, Gue Myung
    JOURNAL OF NONLINEAR AND CONVEX ANALYSIS, 2018, 19 (08) : 1319 - 1329
  • [9] Stochastic Optimization for DC Functions and Non-smooth Non-convex Regularizers with Non-asymptotic Convergence
    Xu, Yi
    Qi, Qi
    Lin, Qihang
    Jin, Rong
    Yang, Tianbao
    INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 97, 2019, 97
  • [10] A method to construct a quasi-normal cone for non-convex and non-smooth set and its applications to non-convex and non-smooth optimization
    Li, Hongwei
    Zhou, Dequn
    Liu, Qinghuai
    WCICA 2006: SIXTH WORLD CONGRESS ON INTELLIGENT CONTROL AND AUTOMATION, VOLS 1-12, CONFERENCE PROCEEDINGS, 2006, : 1585 - +