Graph Representation Learning via Aggregation Enhancement

TL;DR

This paper introduces a kernel regression-based approach to improve information aggregation in graph neural networks, leading to significant performance gains especially for deep models in node classification tasks.

Contribution

It proposes using kernel regression loss to enhance GNN training, addressing aggregation challenges and outperforming state-of-the-art methods in various settings.

Findings

Substantial performance improvements on multiple datasets.

Effective in both self-supervised and supervised learning.

Enhances deep GNNs' capabilities.

Abstract

Graph neural networks (GNNs) have become a powerful tool for processing graph-structured data but still face challenges in effectively aggregating and propagating information between layers, which limits their performance. We tackle this problem with the kernel regression (KR) approach, using KR loss as the primary loss in self-supervised settings or as a regularization term in supervised settings. We show substantial performance improvements compared to state-of-the-art in both scenarios on multiple transductive and inductive node classification datasets, especially for deep networks. As opposed to mutual information (MI), KR loss is convex and easy to estimate in high-dimensional cases, even though it indirectly maximizes the MI between its inputs. Our work highlights the potential of KR to advance the field of graph representation learning and enhance the performance of GNNs. The code to reproduce our experiments is available at https://github.com/Anonymous1252022/KR_for_GNNs