GBK-GNN: Gated Bi-Kernel Graph Neural Networks for Modeling Both Homophily and Heterophily

Du, Lun; Shi, Xiaozhou; Fu, Qiang; Liu, Hengyu; Han, Shi; Zhang, Dongmei

doi:10.1145/3485447.3512201

Cited by 36 publications

(12 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…From the analysis in Sec. 3.1, we find that both the node-level homophily ratio [11,27] and node degree (reflects the recall of nodes from the same class) can affect the performance of GCN on the node classification task, where increasing either of the two variables can lead to better performance of GCN. This finding, i.e., classification performance of GCN on heterophily graphs can be increased by reducing the heterophily-level of graphs, motivates us to design a graph-rewiring strategy to increase homophily-level for heterophily graphs so that GNNs can perform better on the rewired graphs.…”

Section: Introductionmentioning

confidence: 92%

“…These early methods are designed for homophily graphs, and they perform poorly on heterophily graphs. Recently, some studies [1,8,11,27,43] propose to design GNNs for modeling heterophily graphs. MixHop [1] was proposed to aggregate representations from multi-hops neighbors to alleviate heterophily.…”

Section: Related Work 61 Graph Representation Learningmentioning

confidence: 99%

“…GPR-GNN [8] proposed to adaptively learn the Generalized PageRank (GPR) weights to jointly optimize node feature and structural information extraction. More recently, GBK-GNN [11] was designed with bi-kernels for homophilic and heterophilic neighbors respectively.…”

Section: Related Work 61 Graph Representation Learningmentioning

confidence: 99%

“…The aforementioned issues motivate considerable studies around GNNs for heterophily graph. For example, some studies [11,33,37] adjust message passing mechanism for heterophily edges, while others [1,8,27,43] enlarge the receptive field for the message passing. Note that, all these works mitigate the distinguishability issue caused by heterophily from the perspective of the GNN model design.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Make Heterophily Graphs Better Fit GNN: A Graph Rewiring Approach

Bi¹,

Du²,

Fu³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Graph Neural Networks (GNNs) are popular machine learning methods for modeling graph data. A lot of GNNs perform well on homophily graphs while having unsatisfied performance on heterophily graphs. Recently, some researchers turn their attentions to designing GNNs for heterophily graphs by adjusting message passing mechanism or enlarging the receptive field of the message passing. Different from existing works that mitigate the issues of heterophily from model design perspective, we propose to study heterophily graphs from an orthogonal perspective by rewiring the graph structure to reduce heterophily and making the traditional GNNs perform better. Through comprehensive empirical studies and analysis, we verify the potential of the rewiring methods. To fully exploit its potential, we propose a method named Deep Heterophily Graph Rewiring (DHGR) to rewire graphs by adding homophilic edges and pruning heterophilic edges. The detailed way of rewiring is determined by comparing the similarity of label/feature-distribution of node neighbors. Besides, we design a scalable implementation for DHGR to guarantee a high efficiency. DHRG can be easily used as a plug-in module, i.e., a graph pre-processing step, for any GNNs, including both GNN for homophily and heterophily, to boost their performance on the node classification task. To the best of our knowledge, it is the first work studying graph rewiring for heterophily graphs. Extensive experiments on 11 public graph datasets demonstrate the superiority of our proposed methods.

show abstract

Section: Introductionmentioning

confidence: 92%

Section: Related Work 61 Graph Representation Learningmentioning

confidence: 99%

Section: Related Work 61 Graph Representation Learningmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Make Heterophily Graphs Better Fit GNN: A Graph Rewiring Approach

Bi¹,

Du²,

Fu³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…As for heterophilous graphs, the datasets used in most studies dedicated to learning under heterophily are limited to the six graphs adopted by Pei et al (2020): squirrel, chameleon, actor, texas, cornell, and wisconsin. These graphs have become the de-facto standard benchmark for evaluating heterophily-specific models and were used in numerous papers (Zhu et al, 2021;Chien et al, 2021;Yan et al, 2022;Maurya et al, 2022;Li et al, 2022;Wang & Zhang, 2022;Du et al, 2022;Suresh et al, 2021;Bo et al, 2021;Luan et al, 2022;Bodnar et al, 2022). We further discuss these datasets in Section 3.…”

Section: Graph Datasetsmentioning

confidence: 99%

A critical look at the evaluation of GNNs under heterophily: are we really making progress?

Platonov¹,

Kuznedelev²,

Diskin³

et al. 2023

Preprint

View full text Add to dashboard Cite

Node classification is a classical graph representation learning task on which Graph Neural Networks (GNNs) have recently achieved strong results. However, it is often believed that standard GNNs only work well for homophilous graphs, i.e., graphs where edges tend to connect nodes of the same class. Graphs without this property are called heterophilous, and it is typically assumed that specialized methods are required to achieve strong performance on such graphs. In this work, we challenge this assumption. First, we show that the standard datasets used for evaluating heterophily-specific models have serious drawbacks, making results obtained by using them unreliable. The most significant of these drawbacks is the presence of a large number of duplicate nodes in the datasets squirrel and chameleon, which leads to train-test data leakage. We show that removing duplicate nodes strongly affects GNN performance on these datasets. Then, we propose a set of heterophilous graphs of varying properties that we believe can serve as a better benchmark for evaluating the performance of GNNs under heterophily. We show that standard GNNs achieve strong results on these heterophilous graphs, almost always outperforming specialized models. Our datasets and the code for reproducing our experiments are available at https://github.com/yandex-research/heterophilous-graphs.

show abstract