HP-GMN: Graph Memory Networks for Heterophilous Graphs

Junjie Xu; Enyan Dai; Xiang Zhang; Suhang Wang

doi:10.1109/ICDM54844.2022.00165

HP-GMN: Graph Memory Networks for Heterophilous Graphs

Junjie Xu, Enyan Dai, Xiang Zhang, Suhang Wang

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

7 Scopus citations

Abstract

Graph neural networks (GNNs) have achieved great success in various graph problems. However, most GNNs are Message Passing Neural Networks (MPNNs) based on the homophily assumption, where nodes with the same label are connected in graphs. Real-world problems bring us heterophily problems, where nodes with different labels are connected in graphs. MPNNs fail to address the heterophily problem because they mix information from different distributions and are not good at capturing global patterns. Therefore, we investigate a novel Graph Memory Networks model on Heterophilous Graphs (HP-GMN) to the heterophily problem in this paper. In HP-GMN, local information and global patterns are learned by local statistics and the memory to facilitate the prediction. We further propose regularization terms to help the memory learn global information. We conduct extensive experiments to show that our method achieves state-of-the-art performance on both homophilous and heterophilous graphs. The code of this paper can be found at https://github.com/junjie-xu/HP-GMN.

Original language	English (US)
Title of host publication	Proceedings - 22nd IEEE International Conference on Data Mining, ICDM 2022
Editors	Xingquan Zhu, Sanjay Ranka, My T. Thai, Takashi Washio, Xindong Wu
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	1263-1268
Number of pages	6
ISBN (Electronic)	9781665450997
DOIs	https://doi.org/10.1109/ICDM54844.2022.00165
State	Published - 2022
Event	22nd IEEE International Conference on Data Mining, ICDM 2022 - Orlando, United States Duration: Nov 28 2022 → Dec 1 2022

Publication series

Name	Proceedings - IEEE International Conference on Data Mining, ICDM
Volume	2022-November
ISSN (Print)	1550-4786

Conference

Conference	22nd IEEE International Conference on Data Mining, ICDM 2022
Country/Territory	United States
City	Orlando
Period	11/28/22 → 12/1/22

All Science Journal Classification (ASJC) codes

General Engineering

Access to Document

10.1109/ICDM54844.2022.00165

Cite this

Xu, J., Dai, E., Zhang, X., & Wang, S. (2022). HP-GMN: Graph Memory Networks for Heterophilous Graphs. In X. Zhu, S. Ranka, M. T. Thai, T. Washio, & X. Wu (Eds.), Proceedings - 22nd IEEE International Conference on Data Mining, ICDM 2022 (pp. 1263-1268). (Proceedings - IEEE International Conference on Data Mining, ICDM; Vol. 2022-November). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICDM54844.2022.00165

Xu, Junjie ; Dai, Enyan ; Zhang, Xiang et al. / HP-GMN : Graph Memory Networks for Heterophilous Graphs. Proceedings - 22nd IEEE International Conference on Data Mining, ICDM 2022. editor / Xingquan Zhu ; Sanjay Ranka ; My T. Thai ; Takashi Washio ; Xindong Wu. Institute of Electrical and Electronics Engineers Inc., 2022. pp. 1263-1268 (Proceedings - IEEE International Conference on Data Mining, ICDM).

@inproceedings{bc3d398d593844d48c9a3571e78340f1,

title = "HP-GMN: Graph Memory Networks for Heterophilous Graphs",

abstract = "Graph neural networks (GNNs) have achieved great success in various graph problems. However, most GNNs are Message Passing Neural Networks (MPNNs) based on the homophily assumption, where nodes with the same label are connected in graphs. Real-world problems bring us heterophily problems, where nodes with different labels are connected in graphs. MPNNs fail to address the heterophily problem because they mix information from different distributions and are not good at capturing global patterns. Therefore, we investigate a novel Graph Memory Networks model on Heterophilous Graphs (HP-GMN) to the heterophily problem in this paper. In HP-GMN, local information and global patterns are learned by local statistics and the memory to facilitate the prediction. We further propose regularization terms to help the memory learn global information. We conduct extensive experiments to show that our method achieves state-of-the-art performance on both homophilous and heterophilous graphs. The code of this paper can be found at https://github.com/junjie-xu/HP-GMN.",

author = "Junjie Xu and Enyan Dai and Xiang Zhang and Suhang Wang",

note = "Funding Information: ACKNOWLEDGMENT This project was partially supported by NSF projects IIS-1707548, CBET-1638320, IIS-1909702 and IIS-1955851, the Army Research Office under grant W911NF21-1-0198, and DHS CINA under grant E205949D. Publisher Copyright: {\textcopyright} 2022 IEEE.; 22nd IEEE International Conference on Data Mining, ICDM 2022 ; Conference date: 28-11-2022 Through 01-12-2022",

year = "2022",

doi = "10.1109/ICDM54844.2022.00165",

language = "English (US)",

series = "Proceedings - IEEE International Conference on Data Mining, ICDM",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "1263--1268",

editor = "Xingquan Zhu and Sanjay Ranka and Thai, {My T.} and Takashi Washio and Xindong Wu",

booktitle = "Proceedings - 22nd IEEE International Conference on Data Mining, ICDM 2022",

address = "United States",

}

Xu, J, Dai, E, Zhang, X & Wang, S 2022, HP-GMN: Graph Memory Networks for Heterophilous Graphs. in X Zhu, S Ranka, MT Thai, T Washio & X Wu (eds), Proceedings - 22nd IEEE International Conference on Data Mining, ICDM 2022. Proceedings - IEEE International Conference on Data Mining, ICDM, vol. 2022-November, Institute of Electrical and Electronics Engineers Inc., pp. 1263-1268, 22nd IEEE International Conference on Data Mining, ICDM 2022, Orlando, United States, 11/28/22. https://doi.org/10.1109/ICDM54844.2022.00165

HP-GMN: Graph Memory Networks for Heterophilous Graphs. / Xu, Junjie; Dai, Enyan; Zhang, Xiang et al.
Proceedings - 22nd IEEE International Conference on Data Mining, ICDM 2022. ed. / Xingquan Zhu; Sanjay Ranka; My T. Thai; Takashi Washio; Xindong Wu. Institute of Electrical and Electronics Engineers Inc., 2022. p. 1263-1268 (Proceedings - IEEE International Conference on Data Mining, ICDM; Vol. 2022-November).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - HP-GMN

T2 - 22nd IEEE International Conference on Data Mining, ICDM 2022

AU - Xu, Junjie

AU - Dai, Enyan

AU - Zhang, Xiang

AU - Wang, Suhang

N1 - Funding Information: ACKNOWLEDGMENT This project was partially supported by NSF projects IIS-1707548, CBET-1638320, IIS-1909702 and IIS-1955851, the Army Research Office under grant W911NF21-1-0198, and DHS CINA under grant E205949D. Publisher Copyright: © 2022 IEEE.

PY - 2022

Y1 - 2022

N2 - Graph neural networks (GNNs) have achieved great success in various graph problems. However, most GNNs are Message Passing Neural Networks (MPNNs) based on the homophily assumption, where nodes with the same label are connected in graphs. Real-world problems bring us heterophily problems, where nodes with different labels are connected in graphs. MPNNs fail to address the heterophily problem because they mix information from different distributions and are not good at capturing global patterns. Therefore, we investigate a novel Graph Memory Networks model on Heterophilous Graphs (HP-GMN) to the heterophily problem in this paper. In HP-GMN, local information and global patterns are learned by local statistics and the memory to facilitate the prediction. We further propose regularization terms to help the memory learn global information. We conduct extensive experiments to show that our method achieves state-of-the-art performance on both homophilous and heterophilous graphs. The code of this paper can be found at https://github.com/junjie-xu/HP-GMN.

AB - Graph neural networks (GNNs) have achieved great success in various graph problems. However, most GNNs are Message Passing Neural Networks (MPNNs) based on the homophily assumption, where nodes with the same label are connected in graphs. Real-world problems bring us heterophily problems, where nodes with different labels are connected in graphs. MPNNs fail to address the heterophily problem because they mix information from different distributions and are not good at capturing global patterns. Therefore, we investigate a novel Graph Memory Networks model on Heterophilous Graphs (HP-GMN) to the heterophily problem in this paper. In HP-GMN, local information and global patterns are learned by local statistics and the memory to facilitate the prediction. We further propose regularization terms to help the memory learn global information. We conduct extensive experiments to show that our method achieves state-of-the-art performance on both homophilous and heterophilous graphs. The code of this paper can be found at https://github.com/junjie-xu/HP-GMN.

UR - http://www.scopus.com/inward/record.url?scp=85147733018&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85147733018&partnerID=8YFLogxK

U2 - 10.1109/ICDM54844.2022.00165

DO - 10.1109/ICDM54844.2022.00165

M3 - Conference contribution

AN - SCOPUS:85147733018

T3 - Proceedings - IEEE International Conference on Data Mining, ICDM

SP - 1263

EP - 1268

BT - Proceedings - 22nd IEEE International Conference on Data Mining, ICDM 2022

A2 - Zhu, Xingquan

A2 - Ranka, Sanjay

A2 - Thai, My T.

A2 - Washio, Takashi

A2 - Wu, Xindong

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 28 November 2022 through 1 December 2022

ER -

Xu J, Dai E, Zhang X , Wang S. HP-GMN: Graph Memory Networks for Heterophilous Graphs. In Zhu X, Ranka S, Thai MT, Washio T, Wu X, editors, Proceedings - 22nd IEEE International Conference on Data Mining, ICDM 2022. Institute of Electrical and Electronics Engineers Inc. 2022. p. 1263-1268. (Proceedings - IEEE International Conference on Data Mining, ICDM). doi: 10.1109/ICDM54844.2022.00165

HP-GMN: Graph Memory Networks for Heterophilous Graphs

Abstract

Publication series

Conference

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this