Dynamic Graph Segmentation for Deep Graph Neural Networks

被引:1
作者
Kang, Johan Kok Zhi [1 ]
Yang, Suwei [1 ]
Venkatesan, Suriya [2 ]
Tan, Sien Yi [2 ]
Cheng, Feng [2 ]
He, Bingsheng [1 ]
机构
[1] Natl Univ Singapore, Singapore, Singapore
[2] GrabTaxi Holdings, Singapore, Singapore
来源
PROCEEDINGS OF THE 28TH ACM SIGKDD CONFERENCE ON KNOWLEDGE DISCOVERY AND DATA MINING, KDD 2022 | 2022年
关键词
Graph Partitioning; Mixture of Experts; Spatial Concept Drift;
D O I
10.1145/3534678.3539111
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
We present Deep network Dynamic Graph Partitioning (DDGP), a novel algorithm for optimizing the division of large graphs for mixture of expert graph neural networks. Our work is motivated from the observation that real world graphs suffer from spatial concept drift, which is detrimental to neural network training. We answer the question of how we can divide a graph, with vertices in each subgraph sharing a similar distribution, so that an expert network trained over each subgraph may yield the best learning outcome. DDGP is a two pronged algorithm that consists of cluster merging, followed by cluster boundary refinement. We used the training performance of each expert model as feedback to iteratively refine partition boundaries among subgraphs. These partitions are distinct for each model and graph network. We provide theoretical proof of convergence for DDGP boundary refinement as a guarantee for model training stability. Finally, we demonstrate experimentally that DDGP outperforms state-of-the-art graph partitioning algorithms for a regression task on multiple large real world graphs, with GraphSage and Graph Attention as our expert models.
引用
收藏
页码:4601 / 4611
页数:11
相关论文
共 36 条
[1]  
[Anonymous], 2011, EUR S ALG
[2]  
Atwood J, 2016, ADV NEUR IN, V29
[3]   Protein function prediction via graph kernels [J].
Borgwardt, KM ;
Ong, CS ;
Schönauer, S ;
Vishwanathan, SVN ;
Smola, AJ ;
Kriegel, HP .
BIOINFORMATICS, 2005, 21 :I47-I56
[4]   Dynamic Routing Networks [J].
Cai, Shaofeng ;
Shu, Yao ;
Wang, Wei .
2021 IEEE WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION WACV 2021, 2021, :3587-3596
[5]  
Defossez Alexandre, 2020, ARXIV200302395STATML
[6]   Golang-based POI Discovery and Recommendation in Real Time [J].
Fan, Qing ;
Jiao, Lang ;
Dai, Chengcheng ;
Deng, Ziqiang ;
Zhang, Rui .
2019 20TH INTERNATIONAL CONFERENCE ON MOBILE DATA MANAGEMENT (MDM 2019), 2019, :527-532
[7]   Application Driven Graph Partitioning [J].
Fan, Wenfei ;
Jin, Ruochun ;
Liu, Muyang ;
Lu, Ping ;
Luo, Xiaojian ;
Xu, Ruiqi ;
Yin, Qiang ;
Yu, Wenyuan ;
Zhou, Jingren .
SIGMOD'20: PROCEEDINGS OF THE 2020 ACM SIGMOD INTERNATIONAL CONFERENCE ON MANAGEMENT OF DATA, 2020, :1765-1779
[8]   Graph Neural Networks for Social Recommendation [J].
Fan, Wenqi ;
Ma, Yao ;
Li, Qing ;
He, Yuan ;
Zhao, Eric ;
Tang, Jiliang ;
Yin, Dawei .
WEB CONFERENCE 2019: PROCEEDINGS OF THE WORLD WIDE WEB CONFERENCE (WWW 2019), 2019, :417-426
[9]   EFFICIENT ALGORITHMS FOR FINDING MAXIMUM MATCHING IN GRAPHS. [J].
Galil, Zvi .
Computing surveys, 1986, 18 (01) :23-38
[10]  
Gori M., 2005, P 2005 IEEE INT JOIN, DOI DOI 10.1109/IJCNN.2005.1555942