Graph Feature Gating Networks

TL;DR

This paper introduces Graph Feature Gating Networks (GFGN), which enable different feature dimensions in GNNs to contribute heterogeneously during aggregation, improving performance based on spectral embedding insights.

Contribution

The paper proposes a novel GFGN framework with three graph filters that allow for heterogeneous feature dimension contributions, inspired by graph signal denoising theory.

Findings

Effective and robust performance on real-world datasets

Outperforms existing GNNs in various tasks

Demonstrates benefits of dimension-specific aggregation strategies

Abstract

Graph neural networks (GNNs) have received tremendous attention due to their power in learning effective representations for graphs. Most GNNs follow a message-passing scheme where the node representations are updated by aggregating and transforming the information from the neighborhood. Meanwhile, they adopt the same strategy in aggregating the information from different feature dimensions. However, suggested by social dimension theory and spectral embedding, there are potential benefits to treat the dimensions differently during the aggregation process. In this work, we investigate to enable heterogeneous contributions of feature dimensions in GNNs. In particular, we propose a general graph feature gating network (GFGN) based on the graph signal denoising problem and then correspondingly introduce three graph filters under GFGN to allow different levels of contributions from feature dimensions. Extensive experiments on various real-world datasets demonstrate the effectiveness and robustness of the proposed frameworks.