Learning on heterogeneous graphs using high-order relations

A heterogeneous graph consists of different vertices and edges types. Learning on heterogeneous graphs typically employs meta-paths to deal with the heterogeneity by reducing the graph to a homogeneous network, guide random walks or capture semantics. These methods are however sensitive to the choic...

Full description

Saved in:
Bibliographic Details
Main Authors: Lee, See Hian, Ji, Feng, Tay, Wee Peng
Other Authors: School of Electrical and Electronic Engineering
Format: Conference or Workshop Item
Language:English
Published: 2021
Subjects:
Online Access:https://hdl.handle.net/10356/147362
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-147362
record_format dspace
spelling sg-ntu-dr.10356-1473622021-07-07T07:23:27Z Learning on heterogeneous graphs using high-order relations Lee, See Hian Ji, Feng Tay, Wee Peng School of Electrical and Electronic Engineering 2021 International Conference on Acoustics, Speech and Signal Processing (ICASSP) Engineering Graph Neural Network Heterogeneous Graph A heterogeneous graph consists of different vertices and edges types. Learning on heterogeneous graphs typically employs meta-paths to deal with the heterogeneity by reducing the graph to a homogeneous network, guide random walks or capture semantics. These methods are however sensitive to the choice of meta-paths, with suboptimal paths leading to poor performance. In this paper, we propose an approach for learning on heterogeneous graphs without using meta-paths. Specifically, we decompose a heterogeneous graph into different homogeneous relation-type graphs, which are then combined to create higher-order relation-type representations. These representations preserve the heterogeneity of edges and retain their edge directions while capturing the interaction of different vertex types multiple hops apart. This is then complemented with attention mechanisms to distinguish the importance of the relation-type based neighbors and the relation-types themselves. Experiments demonstrate that our model generally outperforms other state-of-the-art baselines in the vertex classification task on three commonly studied heterogeneous graph datasets. Accepted version 2021-07-07T07:23:27Z 2021-07-07T07:23:27Z 2021 Conference Paper Lee, S. H., Ji, F. & Tay, W. P. (2021). Learning on heterogeneous graphs using high-order relations. 2021 International Conference on Acoustics, Speech and Signal Processing (ICASSP). https://hdl.handle.net/10356/147362 en © 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. application/pdf
institution Nanyang Technological University
building NTU Library
continent Asia
country Singapore
Singapore
content_provider NTU Library
collection DR-NTU
language English
topic Engineering
Graph Neural Network
Heterogeneous Graph
spellingShingle Engineering
Graph Neural Network
Heterogeneous Graph
Lee, See Hian
Ji, Feng
Tay, Wee Peng
Learning on heterogeneous graphs using high-order relations
description A heterogeneous graph consists of different vertices and edges types. Learning on heterogeneous graphs typically employs meta-paths to deal with the heterogeneity by reducing the graph to a homogeneous network, guide random walks or capture semantics. These methods are however sensitive to the choice of meta-paths, with suboptimal paths leading to poor performance. In this paper, we propose an approach for learning on heterogeneous graphs without using meta-paths. Specifically, we decompose a heterogeneous graph into different homogeneous relation-type graphs, which are then combined to create higher-order relation-type representations. These representations preserve the heterogeneity of edges and retain their edge directions while capturing the interaction of different vertex types multiple hops apart. This is then complemented with attention mechanisms to distinguish the importance of the relation-type based neighbors and the relation-types themselves. Experiments demonstrate that our model generally outperforms other state-of-the-art baselines in the vertex classification task on three commonly studied heterogeneous graph datasets.
author2 School of Electrical and Electronic Engineering
author_facet School of Electrical and Electronic Engineering
Lee, See Hian
Ji, Feng
Tay, Wee Peng
format Conference or Workshop Item
author Lee, See Hian
Ji, Feng
Tay, Wee Peng
author_sort Lee, See Hian
title Learning on heterogeneous graphs using high-order relations
title_short Learning on heterogeneous graphs using high-order relations
title_full Learning on heterogeneous graphs using high-order relations
title_fullStr Learning on heterogeneous graphs using high-order relations
title_full_unstemmed Learning on heterogeneous graphs using high-order relations
title_sort learning on heterogeneous graphs using high-order relations
publishDate 2021
url https://hdl.handle.net/10356/147362
_version_ 1705151330787524608