Learning Aligned-Spatial Graph Convolutional Networks for Graph Classification

Lu Bai; Yuhang Jiao; Lixin Cui; Edwin R. Hancock

Learning Aligned-Spatial Graph Convolutional Networks for Graph Classification

Lu Bai, Yuhang Jiao, Lixin Cui, Edwin R. Hancock

Computer Science

Research output: Working paper › Preprint

Abstract

In this paper, we develop a novel Aligned-Spatial Graph Convolutional Network (ASGCN) model to learn effective features for graph classification. Our idea is to transform arbitrary-sized graphs into fixed-sized aligned grid structures, and define a new spatial graph convolution operation associated with the grid structures. We show that the proposed ASGCN model not only reduces the problems of information loss and imprecise information representation arising in existing spatially-based Graph Convolutional Network (GCN) models, but also bridges the theoretical gap between traditional Convolutional Neural Network (CNN) models and spatially-based GCN models. Moreover, the proposed ASGCN model can adaptively discriminate the importance between specified vertices during the process of spatial graph convolution, explaining the effectiveness of the proposed model. Experiments on standard graph datasets demonstrate the effectiveness of the proposed model.

Original language	Undefined/Unknown
Publication status	Published - 6 Apr 2019

Publication series

Name	arXiv

Bibliographical note

arXiv admin note: text overlap with arXiv:1902.09936, arXiv:1809.01090

Keywords

cs.LG
stat.ML

Access to Document

https://arxiv.org/abs/1904.04238

Cite this

@techreport{ba94af6f5cce4c67980e9f6cab8c915a,

title = "Learning Aligned-Spatial Graph Convolutional Networks for Graph Classification",

abstract = " In this paper, we develop a novel Aligned-Spatial Graph Convolutional Network (ASGCN) model to learn effective features for graph classification. Our idea is to transform arbitrary-sized graphs into fixed-sized aligned grid structures, and define a new spatial graph convolution operation associated with the grid structures. We show that the proposed ASGCN model not only reduces the problems of information loss and imprecise information representation arising in existing spatially-based Graph Convolutional Network (GCN) models, but also bridges the theoretical gap between traditional Convolutional Neural Network (CNN) models and spatially-based GCN models. Moreover, the proposed ASGCN model can adaptively discriminate the importance between specified vertices during the process of spatial graph convolution, explaining the effectiveness of the proposed model. Experiments on standard graph datasets demonstrate the effectiveness of the proposed model. ",

keywords = "cs.LG, stat.ML",

author = "Lu Bai and Yuhang Jiao and Lixin Cui and Hancock, {Edwin R.}",

note = "arXiv admin note: text overlap with arXiv:1902.09936, arXiv:1809.01090",

year = "2019",

month = apr,

day = "6",

language = "Undefined/Unknown",

series = "arXiv",

type = "WorkingPaper",

}

TY - UNPB

T1 - Learning Aligned-Spatial Graph Convolutional Networks for Graph Classification

AU - Bai, Lu

AU - Jiao, Yuhang

AU - Cui, Lixin

AU - Hancock, Edwin R.

N1 - arXiv admin note: text overlap with arXiv:1902.09936, arXiv:1809.01090

PY - 2019/4/6

Y1 - 2019/4/6

N2 - In this paper, we develop a novel Aligned-Spatial Graph Convolutional Network (ASGCN) model to learn effective features for graph classification. Our idea is to transform arbitrary-sized graphs into fixed-sized aligned grid structures, and define a new spatial graph convolution operation associated with the grid structures. We show that the proposed ASGCN model not only reduces the problems of information loss and imprecise information representation arising in existing spatially-based Graph Convolutional Network (GCN) models, but also bridges the theoretical gap between traditional Convolutional Neural Network (CNN) models and spatially-based GCN models. Moreover, the proposed ASGCN model can adaptively discriminate the importance between specified vertices during the process of spatial graph convolution, explaining the effectiveness of the proposed model. Experiments on standard graph datasets demonstrate the effectiveness of the proposed model.

AB - In this paper, we develop a novel Aligned-Spatial Graph Convolutional Network (ASGCN) model to learn effective features for graph classification. Our idea is to transform arbitrary-sized graphs into fixed-sized aligned grid structures, and define a new spatial graph convolution operation associated with the grid structures. We show that the proposed ASGCN model not only reduces the problems of information loss and imprecise information representation arising in existing spatially-based Graph Convolutional Network (GCN) models, but also bridges the theoretical gap between traditional Convolutional Neural Network (CNN) models and spatially-based GCN models. Moreover, the proposed ASGCN model can adaptively discriminate the importance between specified vertices during the process of spatial graph convolution, explaining the effectiveness of the proposed model. Experiments on standard graph datasets demonstrate the effectiveness of the proposed model.

KW - cs.LG

KW - stat.ML

M3 - Preprint

T3 - arXiv

BT - Learning Aligned-Spatial Graph Convolutional Networks for Graph Classification

ER -