Distance and Hop-wise Structures Encoding Enhanced Graph Attention Networks

TL;DR

This paper introduces a novel method for enhancing Graph Attention Networks by incorporating hop-wise structure and distance information, leading to improved performance in capturing structural features.

Contribution

The work proposes a new approach to embed hop-wise structure and distance information into GATs, which was previously an unexplored idea, improving their ability to capture structural features.

Findings

DHSEGATs achieve competitive results on benchmark tasks.

Embedding structure and distance info improves GNN performance.

The method effectively captures structural features in graphs.

Abstract

Numerous works have proven that existing neighbor-averaging Graph Neural Networks cannot efficiently catch structure features, and many works show that injecting structure, distance, position or spatial features can significantly improve performance of GNNs, however, injecting overall structure and distance into GNNs is an intuitive but remaining untouched idea. In this work, we shed light on the direction. We first extracting hop-wise structure information and compute distance distributional information, gathering with node's intrinsic features, embedding them into same vector space and then adding them up. The derived embedding vectors are then fed into GATs(like GAT, AGDN) and then Correct and Smooth, experiments show that the DHSEGATs achieve competitive result. The code is available at https://github.com/hzg0601/DHSEGATs.