/dp_divergence

Calculation of DP divergence

Primary LanguageC

dp_divergence

Calculation of the DP divergence of two datasets

Requirements

Python 2.7 with numpy, scipy and matplotlib

or Matlab

Install

Matlab version

run

mex generate_graph.c

in Matlab to compile the c code for generating Minimum Spanning Tree

Run test.m

Python version

from dp_div import dp_div

Test code is included in dp_div.py

Examples (Python)

Here are three examples of the divergence between two 2-d Gaussian distributions

small medium large
dp_div=0.101 dp_div=0.572 dp_div=0.999

Citation

If you find this repo is useful, please kindly cite the following papers:

Berisha, V., Wisler, A., Hero, A. O., & Spanias, A. (2016). Empirically estimable classification bounds based on a nonparametric divergence measure. IEEE Transactions on Signal Processing, 64(3), 580-591.

Tu, M., Berisha, V., & Liss, J. (2017, March). Objective assessment of pathological speech using distribution regression. In Acoustics, Speech and Signal Processing (ICASSP), 2017 IEEE International Conference on (pp. 5050-5054). IEEE.