Reducing Oversmoothing through Informed Weight Initialization in Graph Neural Networks

TL;DR

This paper introduces G-Init, a new weight initialization scheme for Graph Neural Networks that reduces oversmoothing, improves deep GNN performance, and is theoretically justified and empirically validated.

Contribution

The paper generalizes Kaiming initialization to GNNs, proposes G-Init to reduce oversmoothing, and provides theoretical analysis and experimental validation.

Findings

G-Init reduces oversmoothing in deep GNNs

Deep GNNs perform well even with no feature information

Theoretical analysis supports improved signal flow in G-Init

Abstract

In this work, we generalize the ideas of Kaiming initialization to Graph Neural Networks (GNNs) and propose a new scheme (G-Init) that reduces oversmoothing, leading to very good results in node and graph classification tasks. GNNs are commonly initialized using methods designed for other types of Neural Networks, overlooking the underlying graph topology. We analyze theoretically the variance of signals flowing forward and gradients flowing backward in the class of convolutional GNNs. We then simplify our analysis to the case of the GCN and propose a new initialization method. Our results indicate that the new method (G-Init) reduces oversmoothing in deep GNNs, facilitating their effective use. Experimental validation supports our theoretical findings, demonstrating the advantages of deep networks in scenarios with no feature information for unlabeled nodes (i.e., ``cold start'' scenario).