Spectral methods are one of the mathematical methods which we analysis networks by computing eigenvalues and eigenvectors. One of the applications of spectral graph theory is Community Detection in social network sciences and Clustering in mathematical concepts. While the spectral graph partitioning methods gives high quality segmentation, we choose this method, but segmenting large graphs by the spectral method is computationally expensive (We need O(n^3) to compute eigenvalues and eigenvectors). So we want to find a method to reduce the cost and use mathematical proofs for our claims. The following methods are popular in spectral methods:

Multi-level Spectral Clustering, Power Iteration Method, Matrix Reordering, Vector Partitioning, Divisive methods such as angle distances, Laplacian Matrix eigenvalues and eigenvectors such as Fiedler Vector

 

Provider

سینا عبداللهی email: sinaabdollahi [at] ut.ac.ir

 

Supervisor

مسعود اسدپور email: asadpour [AT] ut.ac.ir