Sparse Representation on Graphs by Tight Wavelet Frames and Applications

TL;DR

This paper introduces a new framework for tight wavelet frames on manifolds and graphs, enabling efficient sparse representation and processing of graph data for denoising and clustering tasks.

Contribution

It provides a constructive characterization of tight wavelet frames on manifolds and graphs, along with fast transform algorithms and applications to data denoising and semi-supervised clustering.

Findings

WFTG outperforms spectral graph wavelet transform in denoising and clustering.

Proposed models achieve competitive results on real datasets.

Efficient algorithms enable practical application of wavelet frames on graphs.

Abstract

In this paper, we introduce a new (constructive) characterization of tight wavelet frames on non-flat domains in both continuum setting, i.e. on manifolds, and discrete setting, i.e. on graphs; discuss how fast tight wavelet frame transforms can be computed and how they can be effectively used to process graph data. We start with defining the quasi-affine systems on a given manifold $\cM$ that is formed by generalized dilations and shifts of a finite collection of wavelet functions $\Psi:=\{\psi_j: 1\le j\le r\}\subset L_2(\R)$. We further require that $\psi_j$ is generated by some refinable function $\phi$ with mask $a_j$. We present the condition needed for the masks $\{a_j: 0\le j\le r\}$ so that the associated quasi-affine system generated by $\Psi$ is a tight frame for $L_2(\cM)$. Then, we discuss how the transition from the continuum (manifolds) to the discrete setting (graphs) can be naturally done. In order for the proposed discrete tight wavelet frame transforms to be useful in applications, we show how the transforms can be computed efficiently and accurately by proposing the fast tight wavelet frame transforms for graph data (WFTG). Finally, we consider two specific applications of the proposed WFTG: graph data denoising and semi-supervised clustering. Utilizing the sparse representation provided by the WFTG, we propose $\ell_1$-norm based optimization models on graphs for denoising and semi-supervised clustering. On one hand, our numerical results show significant advantage of the WFTG over the spectral graph wavelet transform (SGWT) by [1] for both applications. On the other hand, numerical experiments on two real data sets show that the proposed semi-supervised clustering model using the WFTG is overall competitive with the state-of-the-art methods developed in the literature of high-dimensional data classification, and is superior to some of these methods.