README.md

sWMD

sWMD in python

1.Introduction.

An efficient technique to learn a supervised metric, which we call the Supervised WMD (S-WMD) metric. To see the details, please refer http://papers.nips.cc/paper/6139-graph-clustering-block-models-and-model-free-results.pdf

2.Required.

• Python 2
• numpy
• scipy 0.18.1
• cython

3.Getting start.

Download the fold "dataset" from https://github.com/gaohuang/S-WMD.

Make sure that all the files and documents in a same directory. First compile the code by:

>>> python setup.py build_ext --inplace

Then run the code by:

>>> python swmd.py

For a detailed list of functionality see "functions.pyx"