Another estimation of Laplacian spectrum of the Kronecker product of graphs
Authors:
Milan Bašić,
Branko Arsić,
Zoran Obradović
Abstract:
The relationships between eigenvalues and eigenvectors of a product graph and those of its factor graphs have been known for the standard products, while characterization of Laplacian eigenvalues and eigenvectors of the Kronecker product of graphs using the Laplacian spectra and eigenvectors of the factors turned out to be quite challenging and has remained an open problem to date. Several approac…
▽ More
The relationships between eigenvalues and eigenvectors of a product graph and those of its factor graphs have been known for the standard products, while characterization of Laplacian eigenvalues and eigenvectors of the Kronecker product of graphs using the Laplacian spectra and eigenvectors of the factors turned out to be quite challenging and has remained an open problem to date. Several approaches for the estimation of Laplacian spectrum of the Kronecker product of graphs have been proposed in recent years. However, it turns out that not all the methods are practical to apply in network science models, particularly in the context of multilayer networks. Here we develop a practical and computationally efficient method to estimate Laplacian spectra of this graph product from spectral properties of their factor graphs which is more stable than the alternatives proposed in the literature. We emphasize that a median of the percentage errors of our estimated Laplacian spectrum almost coincides with the $x$-axis, unlike the alternatives which have sudden jumps at the beginning followed by a gradual decrease for the percentage errors. The percentage errors confined (confidence of the estimations) up to $\pm$10% for all considered approximations, depending on a graph density. Moreover, we theoretically prove that the percentage errors becomes smaller when the network grows or the edge density level increases. Additionally, some novel theoretical results considering the exact formulas and lower bounds related to the certain correlation coefficients corresponding to the estimated eigenvectors are presented.
△ Less
Submitted 4 February, 2021;
originally announced February 2021.
Further results on structured regression for multi-scale networks
Authors:
Milan Bašić,
Branko Arsić,
Zoran Obradović
Abstract:
Gaussian Conditional Random Fields (GCRF), as a structured regression model, is designed to achieve higher regression accuracy than unstructured predictors at the expense of execution time, taking into account the objects similarities and the outputs of unstructured predictors simultaneously. As most structural models, the GCRF model does not scale well with large networks. One of the approaches c…
▽ More
Gaussian Conditional Random Fields (GCRF), as a structured regression model, is designed to achieve higher regression accuracy than unstructured predictors at the expense of execution time, taking into account the objects similarities and the outputs of unstructured predictors simultaneously. As most structural models, the GCRF model does not scale well with large networks. One of the approaches consists of performing calculations on factor graphs (if it is possible) rather than on the full graph, which is more computationally efficient. The Kronecker product of the graphs appears to be a natural choice for a graph decomposition. However, this idea is not straightforwardly applicable for GCRF, since characterizing a Laplacian spectrum of the Kronecker product of graphs, which GCRF is based on, from spectra of its factor graphs has remained an open problem. In this paper we apply new estimations for the Laplacian eigenvalues and eigenvectors, and achieve high prediction accuracy of the proposed models, while the computational complexity of the models, compared to the original GCRF model, is improved from $O(n_{1}^{3}n_{2}^{3})$ to $O(n_{1}^{3} + n_{2}^{3})$. Furthermore, we study the GCRF model with a non-Kronecker graph, where the model consists of finding the nearest Kronecker product of graph for an initial graph. Although the proposed models are more complex, they achieve high prediction accuracy too, while the execution time is still much better compare to the original GCRF model. The effectiveness of the proposed models is characterized on three types of random networks where the proposed models were consistently away more accurate than the previously presented GCRF model for multiscale networks [Jesse Glass and Zoran Obradovic. Structured regression on multiscale networks. IEEE Intelligent Systems, 32(2):23-30, 2017.].
△ Less
Submitted 2 September, 2019;
originally announced September 2019.
