Graph neural networks customize node updates for better accuracy
Node-level Graph Neural Architecture Search Framework
Machine LearningArtificial Intelligence
Summary
Graph neural networks typically treat all nodes the same, which can make them less effective and cause problems when networks get deeper. The authors designed a method called N-GNAS that picks the best network setup for different groups of nodes, improving how the network learns. They also use a technique to keep features from different categories distinct, helping the network understand data better. Their approach worked better than existing methods on several tests, including getting an accuracy of 78.26% on a well-known dataset.
What this means in practice
- •For machine learning engineers: Design graph neural networks that adapt update rules to different node groups for improved prediction on structured data.
- •For data analytics teams: Improve analysis workflows for relational data by automatically optimizing graph models to reduce common performance issues.
Authors
Lintao Yanga, Sirui Lia, Yaqing Wang, Pietro Liò, Xu Shen, Baisong Liu, Chengbin Peng
Abstract
In recent years, Graph Neural Networks (GNNs) and architecture search frameworks have gained extensive application in non-Euclidean data processing, attributable to their superior capacity in managing unstructured data. Nevertheless, traditional approaches typically apply uniform convolution operations to all nodes, regardless of their varying structural and feature characteristics, which can undermine model performance and result in over-smoothing issues as the number of layers increases. To overcome this limitation, in this work, we propose a \textbf{N}ode-Level \textbf{G}raph \textbf{N}eural \textbf{A}rchitecture \textbf{S}earch (N-GNAS) algorithm. It can automatically choose an appropriate network architecture for each subset of nodes when updating node features. N-GNAS also introduces a contrastive learning loss to separate sample features from different categories and vice versa. In experiments conducted on eight datasets for node and graph classification, our methodology outperforms current leading GNAS techniques and traditional human-designed GNNs. For example, it achieves an accuracy rate of 78.26\% on the CiteSeer dataset.