Network State Compression

A comparison of three different algorithms for compressing/summarizing a network. These algorithms include:

• Slice Tree: A technique based on the decomposition of the network into smooth regions using slices.

• Spectral Graph Fourier: Extends the notion of the Fourier transform to graphs.

• Spectral Graph Wavelets: Extends the notion of wavelets to graphs using spectral graph theory.

Dependencies

• Python 2.7
• matplotlib
• numpy
• scipy
• networkx
• Graphviz

Cloning

This project uses submodules. To clone all submodules as well, run:

$ git clone --recursive https://github.com/ucsb-igert/nsc.git

Processing

Several datasets are used to benchmark these algorithms. See data/README.md for more information about the datasets.

Two of the datasets are unprocessed. They can be processed with

$ ./processing/process_all

See processing/README.md for more details.

Experiments

After processing the datasets, experiments can be run on them. To run them all, do

$ ./experiments/run_all

See experiments/README.md for more details.