Authors:
Chandramani Chaudhary
1
;
Nirmal Boran
1
;
N. Sangeeth
2
and
Virendra Singh
3
Affiliations:
1
National Institute of Technology Calicut, Kozhikode, India
;
2
National Institute of Technology, Trichy, Tiruchirappalli, India
;
3
Indian Institute of Technology Bombay, Mumbai, India
Keyword(s):
Graph Neural Network, Heterophily, Homophily, Oversmoothing, Decoupling.
Abstract:
By leveraging graph structure, Graph Neural Networks (GNN) have emerged as a useful model for graph-based datasets. While it is widely assumed that GNNs outperform basic neural networks, recent research shows that for some datasets, neural networks outperform GNNs. Heterophily is one of the primary causes of GNN performance degradation, and many models have been proposed to handle it. Furthermore, some intrinsic information in graph structure is often overlooked, such as edge direction. In this work, we propose GNNDLD, a model which exploits the edge direction and label distribution around a node in varying neighborhoods (hop-wise). We combine features from all layers to retain both low-pass frequency and high-pass frequency components of a node because different layers of neural networks provide different types of information. In addition, to avoid oversmoothing, we decouple the node feature aggregation and transformation operations. By combining all of these concepts, we present a
simple yet very efficient model. Experiments on six standard real-world datasets show the superiority of GNNDLD over the state-of-the-art models in both homophily and heterophily.
(More)