Generalized Equivariance and Preferential Labeling for GNN Node Classification

TL;DR

This paper introduces a novel approach for unattributed node classification in graphs, addressing limitations of existing GNNs by proposing generalized equivariance and preferential labeling, leading to improved performance.

Contribution

It proposes a new generalized equivariance property and a preferential labeling method tailored for unattributed node classification in GNNs, overcoming previous limitations.

Findings

Achieves high accuracy on unattributed node classification tasks

Addresses the limitations of existing GNNs with random or uniform labels

Demonstrates the effectiveness of the proposed methods through experiments

Abstract

Existing graph neural networks (GNNs) largely rely on node embeddings, which represent a node as a vector by its identity, type, or content. However, graphs with unattributed nodes widely exist in real-world applications (e.g., anonymized social networks). Previous GNNs either assign random labels to nodes (which introduces artefacts to the GNN) or assign one embedding to all nodes (which fails to explicitly distinguish one node from another). Further, when these GNNs are applied to unattributed node classification problems, they have an undesired equivariance property, which are fundamentally unable to address the data with multiple possible outputs. In this paper, we analyze the limitation of existing approaches to node classification problems. Inspired by our analysis, we propose a generalized equivariance property and a Preferential Labeling technique that satisfies the desired property asymptotically. Experimental results show that we achieve high performance in several unattributed node classification tasks.