Conditional Local Feature Encoding for Graph Neural Networks

TL;DR

This paper introduces Conditional Local Feature Encoding (CLFE), a novel method to preserve node-specific information in GNNs, addressing the issue of feature homogenization in deep message passing, and improves performance across various tasks.

Contribution

The paper proposes CLFE, a new encoding technique that enhances GNNs by maintaining node-specific features during message passing, leading to better performance.

Findings

Consistent performance improvements across seven benchmark datasets.

Effective in four different graph domain tasks.

Enhances baseline GNN models with minimal additional complexity.

Abstract

Graph neural networks (GNNs) have shown great success in learning from graph-based data. The key mechanism of current GNNs is message passing, where a node's feature is updated based on the information passing from its local neighbourhood. A limitation of this mechanism is that node features become increasingly dominated by the information aggregated from the neighbourhood as we use more rounds of message passing. Consequently, as the GNN layers become deeper, adjacent node features tends to be similar, making it more difficult for GNNs to distinguish adjacent nodes, thereby, limiting the performance of GNNs. In this paper, we propose conditional local feature encoding (CLFE) to help prevent the problem of node features being dominated by the information from local neighbourhood. The idea of our method is to extract the node hidden state embedding from message passing process and concatenate it with the nodes feature from previous stage, then we utilise linear transformation to form a CLFE based on the concatenated vector. The CLFE will form the layer output to better preserve node-specific information, thus help to improve the performance of GNN models. To verify the feasibility of our method, we conducted extensive experiments on seven benchmark datasets for four graph domain tasks: super-pixel graph classification, node classification, link prediction, and graph regression. The experimental results consistently demonstrate that our method improves model performance across a variety of baseline GNN models for all four tasks.