Methodology and first-order algorithms for solving nonsmooth and non-strongly convex bilevel optimization problems

被引:3
|
作者
Doron, Lior [1 ]
Shtern, Shimrit [1 ]
机构
[1] Technion Israel Inst Technol, Fac Data & Decis Sci, IL-3200003 Haifa, Israel
关键词
Bilevel optimization; Convex minimization; Complexity analysis; First-order methods; REGRESSION;
D O I
10.1007/s10107-022-01914-4
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
Simple bilevel problems are optimization problems in which we want to find an optimal solution to an inner problem that minimizes an outer objective function. Such problems appear in many machine learning and signal processing applications as a way to eliminate undesirable solutions. In our work, we suggest a new approach that is designed for bilevel problems with simple outer functions, such as the l1 norm, which are not required to be either smooth or strongly convex. In our new ITerative Approximation and Level-set EXpansion (ITALEX) approach, we alternate between expanding the level-set of the outer function and approximately optimizing the inner problem over this level-set. We show that optimizing the inner function through first-order methods such as proximal gradient and generalized conditional gradient results in a feasibility convergence rate of O(1/k), which up to now was a rate only achieved by bilevel algorithms for smooth and strongly convex outer functions. Moreover, we prove an O(1/root k) rate of convergence for the outer function, contrary to existing methods, which only provide asymptotic guarantees. We demonstrate this performance through numerical experiments.
引用
收藏
页码:521 / 558
页数:38
相关论文
共 50 条
  • [31] A global optimization method for solving convex quadratic bilevel programming problems
    Muu, LD
    Quy, NV
    JOURNAL OF GLOBAL OPTIMIZATION, 2003, 26 (02) : 199 - 219
  • [32] A projection-free method for solving convex bilevel optimization problems
    Giang-Tran, Khanh-Hung
    Ho-Nguyen, Nam
    Lee, Dabeen
    MATHEMATICAL PROGRAMMING, 2024,
  • [33] A Global Optimization Method for Solving Convex Quadratic Bilevel Programming Problems
    Le Dung Muu
    Nguyen Van Quy
    Journal of Global Optimization, 2003, 26 : 199 - 219
  • [34] Solving semidefinite quadratic problems within nonsmooth optimization algorithms
    Frangioni, A
    COMPUTERS & OPERATIONS RESEARCH, 1996, 23 (11) : 1099 - 1118
  • [35] First-Order Algorithms Converge Faster than O(1/k) on Convex Problems
    Lee, Ching-pei
    Wright, Stephen J.
    INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 97, 2019, 97
  • [36] A distributed stochastic first-order method for strongly concave-convex saddle point problems
    Qureshi, Muhammad, I
    Khan, Usman A.
    2023 62ND IEEE CONFERENCE ON DECISION AND CONTROL, CDC, 2023, : 4170 - 4175
  • [37] First-Order Algorithms for Robust Optimization Problems via Convex-Concave Saddle-Point Lagrangian Reformulation
    Postek, Krzysztof
    Shtern, Shimrit
    INFORMS JOURNAL ON COMPUTING, 2024,
  • [38] High-Resolution Modeling of the Fastest First-Order Optimization Method for Strongly Convex Functions
    Sun, Boya
    George, Jemin
    Kia, Solmaz
    2020 59TH IEEE CONFERENCE ON DECISION AND CONTROL (CDC), 2020, : 4237 - 4242
  • [39] Practical First-Order Bayesian Optimization Algorithms
    Prakash, Utkarsh
    Khatwani, Kushagra
    Chollera, Aryan Gautam
    Prabuchandran, K. J.
    Bodas, Tejas
    PROCEEDINGS OF 7TH JOINT INTERNATIONAL CONFERENCE ON DATA SCIENCE AND MANAGEMENT OF DATA, CODS-COMAD 2024, 2024, : 173 - 181
  • [40] On the rate of convergence of alternating minimization for non-smooth non-strongly convex optimization in Banach spaces
    Both, Jakub Wiktor
    OPTIMIZATION LETTERS, 2022, 16 (02) : 729 - 743