Convergence of a first-order consensus-based global optimization algorithm

被引:44
作者
Ha, Seung-Yeal [1 ,2 ,3 ]
Jin, Shi [4 ,5 ]
Kim, Doheon [3 ]
机构
[1] Seoul Natl Univ, Dept Math Sci, Seoul 08826, South Korea
[2] Seoul Natl Univ, Res Inst Math, Seoul 08826, South Korea
[3] Korea Inst Adv Study, Sch Math, Hoegiro 85, Seoul 02455, South Korea
[4] Shanghai Jiao Tong Univ, Sch Math Sci, Shanghai 200240, Peoples R China
[5] Shanghai Jiao Tong Univ, Inst Nat Sci, Shanghai 200240, Peoples R China
基金
新加坡国家研究基金会;
关键词
Consensus-based optimization; Gibb’ s distribution; global optimization; machine learning; objective function; FLOCKING; DYNAMICS; MODEL;
D O I
10.1142/S0218202520500463
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
Global optimization of a non-convex objective function often appears in large-scale machine learning and artificial intelligence applications. Recently, consensus-based optimization (CBO) methods have been introduced as one of the gradient-free optimization methods. In this paper, we provide a convergence analysis for the first-order CBO method in [J. A. Carrillo, S. Jin, L. Li and Y. Zhu, A consensus-based global optimization method for high dimensional machine learning problems, https://arxiv.org/abs/1909.09249v1]. Prior to this work, the convergence study was carried out for CBO methods on corresponding mean-field limit, a Fokker-Planck equation, which does not imply the convergence of the CBO method per se. Based on the consensus estimate directly on the first-order CBO model, we provide a convergence analysis of the first-order CBO method [J. A. Carrillo, S. Jin, L. Li and Y. Zhu, A consensus-based global optimization method for high dimensional machine learning problems, https://arxiv.org/abs/1909.09249v1] without resorting to the corresponding mean-field model. Our convergence analysis consists of two steps. In the first step, we show that the CBO model exhibits a global consensus time asymptotically for any initial data, and in the second step, we provide a sufficient condition on system parameters - which is dimension independent - and initial data which guarantee that the converged consensus state lies in a small neighborhood of the global minimum almost surely.
引用
收藏
页码:2417 / 2444
页数:28
相关论文
共 31 条
[11]   Emergent behaviors of the Cucker-Smale ensemble under attractive-repulsive couplings and Rayleigh frictions [J].
Fang, Di ;
Ha, Seung-Yeal ;
Jin, Shi .
MATHEMATICAL MODELS & METHODS IN APPLIED SCIENCES, 2019, 29 (07) :1349-1385
[12]  
Ha S.-Y., CONVERGENCE ERROR ES
[13]  
Ha SY, 2009, COMMUN MATH SCI, V7, P453
[14]  
Ha SY, 2009, COMMUN MATH SCI, V7, P297
[15]   GENETIC ALGORITHMS [J].
HOLLAND, JH .
SCIENTIFIC AMERICAN, 1992, 267 (01) :66-72
[16]   Random Batch Methods (RBM) for interacting particle systems [J].
Jin, Shi ;
Li, Lei ;
Liu, Jian-Guo .
JOURNAL OF COMPUTATIONAL PHYSICS, 2020, 400
[17]  
Kennedy J, 1995, 1995 IEEE INTERNATIONAL CONFERENCE ON NEURAL NETWORKS PROCEEDINGS, VOLS 1-6, P1942, DOI 10.1109/icnn.1995.488968
[18]  
Kennedy J., 2006, HDB NATUREINSPIRED I, p187, DOI DOI 10.1007/0-387-27705-6_6
[19]   OPTIMIZATION BY SIMULATED ANNEALING [J].
KIRKPATRICK, S ;
GELATT, CD ;
VECCHI, MP .
SCIENCE, 1983, 220 (4598) :671-680
[20]   Emergent behaviour in multi-particle systems with non-local interactions [J].
Kolokolnikov, Theodore ;
Carrillo, Jose A. ;
Bertozzi, Andrea ;
Fetecau, Razvan ;
Lewis, Mark .
PHYSICA D-NONLINEAR PHENOMENA, 2013, 260 :1-4