CS-Over-Graphs

[AIP Project] This repository contains the implementation of our course project for Advanced Image Processing [CS754] : Compressed Sensing Over Graph Structures

In this project we try to explore the utility of Compressive Sensing to identify the central nodes of information flow (i.e nodes with the highest betweenness centrality). We use the following datasets: Facebook and GNUTELLA for studying the viability/performance of the DICeNod Algorithm as defined in the paper.

Network Plots with the Top 10 central nodes

Size of the nodes gives an idea about the amount of information flow, through it.

Facebook Network(Large Graph Layout) with labelled Top 10 nodes with highest information flow

GNUTELLA Network(Kamada Kawai) with labelled Top 10 nodes with highest information flow

We also studied the change in reconstruction error with the increase in the number of measurements and other parameters.

You can find rest of the plots/results/mathematical formulations/theoretical guarantees in the Project Report. You can check out the implementation in DICeNOD sub-directory.

Authors

Name	Contact
Richeek Das	richeek@cse.iitb.ac.in
Aaron Jerry Ninan	190100001@iitb.ac.in

May 2021.