By Topic

Canonical Correlation Analysis for Detecting Changes in Network Structure

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$31 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

3 Author(s)
O'Sullivan, A. ; Dept. Of Math., Imperial Coll. London, London, UK ; Adams, N.M. ; Rezek, I.

Methods for the analysis of network data can be divided into two approaches, there are structural pattern recognition methods that act directly upon the relational data and alternatively the application of well studied standard statistical techniques which first require a transformation of the data to a vector representation. There are a number of methods available for the comparison of networks within the structural field, for example Frequent Sub graph Mining can be used to classify a collection of graphs into classes and spectral decomposition allows for anomaly detection. In this paper we detour from the standard structural methods and instead propose a novel combination of network embedding and Canonical Correlation Analysis(CCA) that allows comparison of coherent networks. We construct a test based on CCA that allows us to detect statistically significant differences between two graphs. This method is demonstrated on a number of simulated networks and also on the VAST Challenge 2008 cell phone call records data. These experiments suggest that the method is well suited for comparing networks of different types and hence is a new unsupervised method for graph comparison that does not look for specific changes in any one feature of a graph.

Published in:

Data Mining Workshops (ICDMW), 2012 IEEE 12th International Conference on

Date of Conference:

10-10 Dec. 2012