Bridging Smoothness and Approximation: Theoretical Insights into Over-Smoothing in Graph Neural Networks
Guangrui Yang, Jianfei Li, Ming Li, Han Feng, Ding-Xuan Zhou
TL;DR
This work develops a graph-based approximation theory by introducing a $K$-functional on graphs and proving its strong equivalence to the modulus of smoothness, enabling quantitative analysis of GCNs' approximation ability under over-smoothing. It shows that high-frequency energy decays exponentially with network depth for common graph filters and establishes a lower bound on the approximation error in terms of smoothness measures, linking network dynamics to function smoothness. The authors validate the theory with experiments on SBM graphs, demonstrating energy decay patterns and highlight that skip-connection variants seperti APPNP and GCNII can mitigate smoothing effects by better preserving high-frequency information. Overall, the results provide a principled lens for understanding GCN limitations and offer architectural guidance to balance frequency components for improved graph function approximation.
Abstract
In this paper, we explore the approximation theory of functions defined on graphs. Our study builds upon the approximation results derived from the $K$-functional. We establish a theoretical framework to assess the lower bounds of approximation for target functions using Graph Convolutional Networks (GCNs) and examine the over-smoothing phenomenon commonly observed in these networks. Initially, we introduce the concept of a $K$-functional on graphs, establishing its equivalence to the modulus of smoothness. We then analyze a typical type of GCN to demonstrate how the high-frequency energy of the output decays, an indicator of over-smoothing. This analysis provides theoretical insights into the nature of over-smoothing within GCNs. Furthermore, we establish a lower bound for the approximation of target functions by GCNs, which is governed by the modulus of smoothness of these functions. This finding offers a new perspective on the approximation capabilities of GCNs. In our numerical experiments, we analyze several widely applied GCNs and observe the phenomenon of energy decay. These observations corroborate our theoretical results on exponential decay order.
