Generalized Laplacian Regularized Framelet Graph Neural Networks

TL;DR

This paper proposes a novel p-Laplacian based Framelet Graph Neural Network approach that enhances multi-resolution graph signal processing, demonstrating superior performance in node classification and signal denoising tasks.

Contribution

It introduces two new models, pL-UFG and pL-fUFG, combining p-Laplacian with framelet techniques for improved graph signal analysis.

Findings

Excellent performance in node classification

Effective in signal denoising

Outperforms existing methods in experiments

Abstract

This paper introduces a novel Framelet Graph approach based on p-Laplacian GNN. The proposed two models, named p-Laplacian undecimated framelet graph convolution (pL-UFG) and generalized p-Laplacian undecimated framelet graph convolution (pL-fUFG) inherit the nature of p-Laplacian with the expressive power of multi-resolution decomposition of graph signals. The empirical study highlights the excellent performance of the pL-UFG and pL-fUFG in different graph learning tasks including node classification and signal denoising.