Vybrané aplikace metody spektrálních rozkladů

Abstract

The master thesis deals with the algorithms that use spectral decompositions of Laplacian matrices. The aim of the thesis is to examine, implement and use these algorithms in various areas. Three spectral clustering algorithms were chosen and used for the analysis of different datasets containing undirected weighted graphs as well as for the image segmentation. The results of clustering analysis of graphs datasets were evaluated by methods measuring the quality of clustering and they were also visualised by Gephi software.

Description

Import 03/11/2016

Subject(s)

master thesis, clustering, cluster, spectral clustering, eigenvalues, eigenvectors, Alglib, Math.NET Numerics, C#, WPF, data analysis, k-means, modularity, conductance, cutratio, silhouette index, dunn index, graph theory, Gephi, programming language R, image segmentation

Citation