Softwares in My Research

   

Probability of Success Plot for Sparse KLIEP (NEW)

Run demo_POS then plot_POS.
demo_POS takes about 2 hours to run on a 16-core Xeon machine using parfor.

MATLAB code

700 

Sparse KLIEP (NEW)

Learning sparse changes between two MNs where two sets of samples are drawn. This implementation can process 500 dimensional MNs using a moderate performance laptop withing 10 minutes.

Demo Page
MATLAB code C Plus Plus code

Fast and Simple Log-linear KLIEP (NEW)

Learning the density ratio function between two densities where two sets of samples are drawn.

Demo Page
MATLAB code

Structural Change Detection (OLD)

We detect structural changes between Markov Networks by density ratio estimation.

Demo Page

Change-Point Detection by Relative Density Ratio Estimation

We detect changes using the Relative Density Ratio estimated by Relative unconstrained Least-Squares Importance Fitting (RuLSIF) method.

Code

Relative Density Ratio Estimation (RuLSIF) by Python

This is a Python implementation of Relative Density Ratio Estimation (RuLSIF) method, whose detail can be found here. With the support of pylab, this software implements the exact same algorithm described in the above webpage.

Demo Page

Song Liu (song@sg.cs.titech.ac.jp)