• Complex
  • Title
  • Keyword
  • Abstract
  • Scholars
  • Journal
  • ISSN
  • Conference
搜索

Author:

Du, Lun (Du, Lun.) | Shi, Xiaozhou (Shi, Xiaozhou.) | Fu, Qiang (Fu, Qiang.) | Ma, Xiaojun (Ma, Xiaojun.) | Liu, Hengyu (Liu, Hengyu.) | Han, Shi (Han, Shi.) | Zhang, Dongmei (Zhang, Dongmei.)

Indexed by:

CPCI-S EI Scopus

Abstract:

Graph Neural Networks (GNNs) are widely used on a variety of graph-based machine learning tasks. For node-level tasks, GNNs have strong power to model the homophily property of graphs (i.e., connected nodes are more similar), while their ability to capture heterophily property is often doubtful. This is partially caused by the design of the feature transformation with the same kernel for the nodes in the same hop and the followed aggregation operator. One kernel cannot model the similarity and the dissimilarity (i.e., the positive and negative correlation) between node features simultaneously even though we use attention mechanisms like Graph Attention Network (GAT), since the weight calculated by attention is always a positive value. In this paper, we propose a novel GNN model based on a bi-kernel feature transformation and a selection gate. Two kernels capture homophily and heterophily information respectively, and the gate is introduced to select which kernel we should use for the given node pairs. We conduct extensive experiments on various datasets with different homophily-heterophily properties. The experimental results show consistent and significant improvements against state-of-the-art GNN methods.

Keyword:

graph neural networks graph mining heterophily homophily

Author Community:

  • [ 1 ] [Du, Lun]Microsoft Res Asia, Beijing, Peoples R China
  • [ 2 ] [Fu, Qiang]Microsoft Res Asia, Beijing, Peoples R China
  • [ 3 ] [Han, Shi]Microsoft Res Asia, Beijing, Peoples R China
  • [ 4 ] [Zhang, Dongmei]Microsoft Res Asia, Beijing, Peoples R China
  • [ 5 ] [Shi, Xiaozhou]Beijing Univ Technol, Beijing, Peoples R China
  • [ 6 ] [Ma, Xiaojun]Peking Univ, Beijing, Peoples R China
  • [ 7 ] [Liu, Hengyu]Northeastern Univ, Shenyang, Peoples R China
  • [ 8 ] [Shi, Xiaozhou]MSRA, Beijing, Peoples R China
  • [ 9 ] [Fu, Qiang]MSRA, Beijing, Peoples R China

Reprint Author's Address:

Show more details

Related Keywords:

Related Article:

Source :

PROCEEDINGS OF THE ACM WEB CONFERENCE 2022 (WWW'22)

Year: 2022

Page: 1550-1558

Cited Count:

WoS CC Cited Count: 53

SCOPUS Cited Count: 87

ESI Highly Cited Papers on the List: 0 Unfold All

WanFang Cited Count:

Chinese Cited Count:

30 Days PV: 1

Affiliated Colleges:

Online/Total:767/5279928
Address:BJUT Library(100 Pingleyuan,Chaoyang District,Beijing 100124, China Post Code:100124) Contact Us:010-67392185
Copyright:BJUT Library Technical Support:Beijing Aegean Software Co., Ltd.