Abstract:Graph structures have proven computationally cumbersome for pattern analysis. The reason for this is that, before graphs can be converted to pattern vectors, correspondences must be established between the nodes of structures which are potentially of different size. To overcome this problem, in this paper, we turn to the spectral decomposition of the Laplacian matrix. We show how the elements of the spectral matrix for the Laplacian can be used to construct symmetric polynomials that are permutation invariants. The coefficients of these polynomials can be used as graph features which can be encoded in a vectorial manner. We extend this representation to graphs in which there are unary attributes on the nodes and binary attributes on the edges by using the spectral decomposition of a Hermitian property matrix that can be viewed as a complex analogue of the Laplacian. To embed the graphs in a pattern space, we explore whether the vectors of invariants can be embedded in a low-dimensional space using a number of alternative strategies, including principal components analysis (PCA), multidimensional scaling (MDS), and locality preserving projection (LPP). Experimentally, we demonstrate that the embeddings result in well-defined graph clusters. Our experiments with the spectral representation involve both synthetic and real-world data. The experiments with synthetic data demonstrate that the distances between spectral feature vectors can be used to discriminate between graphs on the basis of their structure. The real-world experiments show that the method can be used to locate clusters of graphs.

Geometric structure of graph Laplacian embeddings

The geometry of kernelized spectral clustering

Spectral Convergence Rate of Graph Laplacian

Spectral clustering of graphs

Geometric Characterisation of Graphs

Structural Analysis of Laplacian Spectral Properties of Large-Scale Networks

Spectral embedding and the latent geometry of multipartite networks

Laplacian Mixture Modeling for Network Analysis and Unsupervised Learning on Graphs

Image Clustering Based on Spectral Features of Laplacian Graph

Graph Manifolds from Spectral Polynomials

A spectral generative model for graph structure

Graph Laplacians, Riemannian Manifolds and Their Machine-Learning

Spectral clustering in the Gaussian mixture block model

Connecting Dots -- from Local Covariance to Empirical Intrinsic Geometry and Locally Linear Embedding

Pattern Vectors from Algebraic Graph Theory

Graph Spectral Decomposition and Clustering

Spectral embedding of graphs

On the Eigenvectors of P-Laplacian

Laplacian Spectral Properties of Graphs from Random Local Samples

Spectral Clustering via sparse graph structure learning with application to Proteomic Signaling Networks in Cancer