Learning Efficient Hash Codes for Fast Graph-Based Data Similarity Retrieval

被引:4
作者
Wang, Jinbao [1 ]
Xu, Shuo [2 ]
Zheng, Feng [1 ]
Lu, Ke [3 ,4 ]
Song, Jingkuan [5 ]
Shao, Ling [6 ]
机构
[1] Southern Univ Sci & Technol, Dept Comp Sci & Engn, Shenzhen 518055, Peoples R China
[2] Anhui Univ, Sch Elect & Informat Engn, Hefei 230039, Peoples R China
[3] Univ Chinese Acad Sci, Sch Engn Sci, Beijing 100049, Peoples R China
[4] Peng Cheng Lab, Shenzhen 518055, Peoples R China
[5] Univ Elect Sci & Technol China, Sch Comp Sci & Engn, Chengdu 610054, Peoples R China
[6] Incept Inst Artificial Intelligence, Abu Dhabi, U Arab Emirates
基金
中国国家自然科学基金;
关键词
Task analysis; Measurement; Computational modeling; Binary codes; Graph neural networks; Data models; Training; Graph representation; graph neural networks; hash codes; similarity retrieval; EDIT DISTANCE; QUANTIZATION;
D O I
10.1109/TIP.2021.3093387
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Traditional operations, e.g. graph edit distance (GED), are no longer suitable for processing the massive quantities of graph-structured data now available, due to their irregular structures and high computational complexities. With the advent of graph neural networks (GNNs), the problems of graph representation and graph similarity search have drawn particular attention in the field of computer vision. However, GNNs have been less studied for efficient and fast retrieval after graph representation. To represent graph-based data, and maintain fast retrieval while doing so, we introduce an efficient hash model with graph neural networks (HGNN) for a newly designed task (i.e. fast graph-based data retrieval). Due to its flexibility, HGNN can be implemented in both an unsupervised and supervised manner. Specifically, by adopting a graph neural network and hash learning algorithms, HGNN can effectively learn a similarity-preserving graph representation and compute pair-wise similarity or provide classification via low-dimensional compact hash codes. To the best of our knowledge, our model is the first to address graph hashing representation in the Hamming space. Our experimental results reach comparable prediction accuracy to full-precision methods and can even outperform traditional models in some cases. In real-world applications, using hash codes can greatly benefit systems with smaller memory capacities and accelerate the retrieval speed of graph-structured data. Hence, we believe the proposed HGNN has great potential in further research.
引用
收藏
页码:6321 / 6334
页数:14
相关论文
共 58 条
[1]   VQA: Visual Question Answering [J].
Agrawal, Aishwarya ;
Lu, Jiasen ;
Antol, Stanislaw ;
Mitchell, Margaret ;
Zitnick, C. Lawrence ;
Parikh, Devi ;
Batra, Dhruv .
INTERNATIONAL JOURNAL OF COMPUTER VISION, 2017, 123 (01) :4-31
[2]  
Amer M. R., 2019, ARXIV190502850
[3]  
[Anonymous], 2019, ARXIV190412787
[4]  
[Anonymous], 2018, INT C MACH LEARN
[5]  
[Anonymous], 2009, NEURIPS, P1753
[6]  
[Anonymous], 2009, NEURIPS
[7]   SimGNN: A Neural Network Approach to Fast Graph Similarity Computation [J].
Bai, Yunsheng ;
Ding, Hao ;
Bian, Song ;
Chen, Ting ;
Sun, Yizhou ;
Wang, Wei .
PROCEEDINGS OF THE TWELFTH ACM INTERNATIONAL CONFERENCE ON WEB SEARCH AND DATA MINING (WSDM'19), 2019, :384-392
[8]  
Bresson Xavier, 2017, ARXIV171107553
[9]   Geometric Deep Learning Going beyond Euclidean data [J].
Bronstein, Michael M. ;
Bruna, Joan ;
LeCun, Yann ;
Szlam, Arthur ;
Vandergheynst, Pierre .
IEEE SIGNAL PROCESSING MAGAZINE, 2017, 34 (04) :18-42
[10]   A graph distance metric based on the maximal common subgraph [J].
Bunke, H ;
Shearer, K .
PATTERN RECOGNITION LETTERS, 1998, 19 (3-4) :255-259