Distributed Optimization for Graph Matching

被引:4
作者
Van Tran, Quoc [1 ]
Sun, Zhiyong [2 ]
Anderson, Brian D. O. [3 ]
Ahn, Hyo-Sung [1 ]
机构
[1] Gwangju Inst Sci & Technol, Sch Mech Engn, Gwangju 61005, South Korea
[2] Eindhoven Univ Technol, Dept Elect Engn, NL-5612 AZ Eindhoven, Netherlands
[3] Australian Natl Univ, Res Sch Elect Energy & Mat Engn, Acton, ACT 2601, Australia
基金
澳大利亚研究理事会; 新加坡国家研究基金会;
关键词
Optimization; Nonhomogeneous media; Task analysis; Labeling; Sun; Digital twin; Convergence; Cyber-physical systems; distributed optimization; graph matching (GM); ALGORITHM; SYSTEM;
D O I
10.1109/TCYB.2022.3140338
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Graph matching, or the determination of the vertex correspondences between a pair of graphs, is a crucial task in various problems in different science and engineering disciplines. This article aims to propose a distributed optimization approach for graph matching (GM) between two isomorphic graphs over multiagent networks. For this, we first show that for a class of asymmetric graphs, GM of two isomorphic graphs is equivalent to a convex relaxation where the set of permutation matrices is replaced by the set of pseudostochastic matrices. Then, we formulate GM as a distributed convex optimization problem with equality constraints and a set constraint, over a network of multiple agents. For arbitrary labelings of the vertices, each agent only has information about just one vertex and its neighborhood, and can exchange information with its neighbors. A projected primal-dual gradient method is developed to solve the constrained optimization problem, and globally exponential convergence of the agents' states to the optimal permutation is achieved. Finally, we illustrate the effectiveness of the algorithm through simulation examples.
引用
收藏
页码:4815 / 4828
页数:14
相关论文
共 46 条
[1]   On convex relaxation of graph isomorphism [J].
Aflalo, Yonathan ;
Bronstein, Alexander ;
Kimmel, Ron .
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2015, 112 (10) :2942-2947
[2]   Distributed Coordination for Optimal Energy Generation and Distribution in Cyber-Physical Energy Networks [J].
Ahn, Hyo-Sung ;
Kim, Byeong-Yeon ;
Lim, Young-Hun ;
Lee, Byung-Hun ;
Oh, Kwang-Kyo .
IEEE TRANSACTIONS ON CYBERNETICS, 2018, 48 (03) :941-954
[3]   DECENTRALIZED GRADIENT ALGORITHM FOR SOLUTION OF A LINEAR EQUATION [J].
Anderson, Brian D. O. ;
Mou, Shaoshuai ;
Morse, A. Stephen ;
Helmke, Uwe .
NUMERICAL ALGEBRA CONTROL AND OPTIMIZATION, 2016, 6 (03) :319-328
[4]  
Boyd S., 2011, Distributed Optimization and Statistical Learning Via the Alternating Direction Method of Multipliers, DOI DOI 10.1561/2200000016
[5]   An eigenspace projection clustering method for inexact graph matching [J].
Caelli, T ;
Kosinov, S .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2004, 26 (04) :515-519
[6]  
Caron R., 2005, Technical report, DOI DOI 10.13140/RG.2.1.4432.8169
[7]   Using a graph-based data mining system to perform web search [J].
Cook, DJ ;
Manocha, N ;
Holder, LB .
INTERNATIONAL JOURNAL OF PATTERN RECOGNITION AND ARTIFICIAL INTELLIGENCE, 2003, 17 (05) :705-720
[8]   A robust alignment-free fingerprint hashing algorithm based on minimum distance graphs [J].
Das, Priyanka ;
Karthik, Kannan ;
Garai, Boul Chandra .
PATTERN RECOGNITION, 2012, 45 (09) :3373-3388
[9]  
Deng W., 2019, ARXIV191108760
[10]   A Tensor-Based Algorithm for High-Order Graph Matching [J].
Duchenne, Olivier ;
Bach, Francis ;
Kweon, In-So ;
Ponce, Jean .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2011, 33 (12) :2383-2395