Ensemble Quadratic Assignment Network for Graph Matching

被引:1
|
作者
Tan, Haoru [1 ,2 ]
Wang, Chuang [1 ,2 ]
Wu, Sitong [2 ]
Zhang, Xu-Yao [1 ,2 ]
Yin, Fei [1 ,2 ]
Liu, Cheng-Lin [1 ,2 ]
机构
[1] Chinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100190, Peoples R China
[2] Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing 100044, Peoples R China
基金
中国国家自然科学基金;
关键词
Graph matching; Combinatorial optimization; Graph neural network; Ensemble learning; RECOGNITION; TRACKING;
D O I
10.1007/s11263-024-02040-8
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Graph matching is a commonly used technique in computer vision and pattern recognition. Recent data-driven approaches have improved the graph matching accuracy remarkably, whereas some traditional algorithm-based methods are more robust to feature noises, outlier nodes, and global transformation (e.g. rotation). In this paper, we propose a graph neural network (GNN) based approach to combine the advantage of data-driven and traditional methods. In the GNN framework, we transform traditional graph matching solvers as single-channel GNNs on the association graph and extend the single-channel architecture to the multi-channel network. The proposed model can be seen as an ensemble method that fuses multiple algorithms at every iteration. Instead of averaging the estimates at the end of the ensemble, in our approach, the independent iterations of the ensembled algorithms exchange their information after each iteration via a 1x1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1\,\times \,1$$\end{document} channel-wise convolution layer. Experiments show that our model improves the performance of traditional algorithms significantly. In addition, we propose a random sampling strategy to reduce the computational complexity and GPU memory usage, so that the model is applicable to matching graphs with thousands of nodes. We evaluate the performance of our method on three tasks: geometric graph matching, semantic feature matching, and few-shot 3D shape classification. The proposed model performs comparably or outperforms the best existing GNN-based methods.
引用
收藏
页码:3633 / 3655
页数:23
相关论文
共 50 条
  • [31] Sparse graph matching network for temporal language localization in videos
    Wu, Guangli
    Xu, Tongjie
    Zhang, Jing
    COMPUTER VISION AND IMAGE UNDERSTANDING, 2024, 240
  • [32] Natalie 2.0: Sparse Global Network Alignment as a Special Case of Quadratic Assignment
    El-Kebir, Mohammed
    Heringa, Jaap
    Klau, Gunnar W.
    ALGORITHMS, 2015, 8 (04) : 1035 - 1051
  • [33] Quadratic Convex Reformulation for Solving Task Assignment Problem with Continuous Hopfield Network
    Hami, Youssef
    Loqman, Chakir
    INTERNATIONAL JOURNAL OF COMPUTATIONAL INTELLIGENCE AND APPLICATIONS, 2021, 20 (04)
  • [34] Linearizing Quadratic Assignment Problem
    Singh, Surya Prakash
    20TH INTERNATIONAL CONFERENCE, EURO MINI CONFERENCE CONTINUOUS OPTIMIZATION AND KNOWLEDGE-BASED TECHNOLOGIES, EUROPT'2008, 2008, : 25 - 30
  • [35] Graph matching based on local and global information of the graph nodes
    Zhan, Yaru
    Zhao, Xiuyang
    Lin, Xue
    Liu, Junkai
    Liu, Mingjun
    Niu, Dongmei
    MULTIMEDIA TOOLS AND APPLICATIONS, 2020, 79 (17-18) : 11567 - 11590
  • [36] A survey for the quadratic assignment problem
    Loiola, Eliane Maria
    de Abreu, Nair Maria Maia
    Boaventura-Netto, Paulo Oswaldo
    Hahn, Peter
    Querido, Tania
    EUROPEAN JOURNAL OF OPERATIONAL RESEARCH, 2007, 176 (02) : 657 - 690
  • [37] Quadratic and multidimensional assignment problems
    Pardalos, PM
    Pitsoulis, LS
    NONLINEAR OPTIMIZATION AND RELATED TOPICS, 2000, 36 : 235 - 256
  • [38] A graph-matching based intra-node task assignment methodology for SMP clusters
    Gao, HE
    Schmidt, A
    Gupta, A
    Luksch, P
    7TH WORLD MULTICONFERENCE ON SYSTEMICS, CYBERNETICS AND INFORMATICS, VOL II, PROCEEDINGS: COMPUTER SCIENCE AND ENGINEERING, 2003, : 344 - 349
  • [39] Cerebrovascular network registration via an efficient attributed graph matching technique
    Almasi, Sepideh
    Lauric, Alexandra
    Malek, Adel
    Miller, Eric L.
    MEDICAL IMAGE ANALYSIS, 2018, 46 : 118 - 129
  • [40] Cross Attention Graph Matching Network for Image-Text Retrieval
    Yang, Xiaoyu
    Xie, Hao
    Mao, Junyi
    Wang, Zhiguo
    Yin, Guangqiang
    PROCEEDINGS OF THE 13TH INTERNATIONAL CONFERENCE ON COMPUTER ENGINEERING AND NETWORKS, VOL II, CENET 2023, 2024, 1126 : 274 - 286