KullbLeiblKLD: Kullback-Leibler divergence

In Mthrun/Distances: Functions For Different Distance Measures

Description

Kullback-Leibler divergence KLD of discrete distributions

Usage

KullbLeiblKLD(P,Q,sym=1)

Arguments

P	two distributions having the same dirccrete range unique([P,Q])
Q	two distributions having the same dirccrete range unique([P,Q])
sym	default ==1: if sym ==1 then the symmetical KLD is returned

Details

NaN in the distributions are ignored. not defined for zero probabilities, replace with very small values,

Value

KLD	Kullback-Leibler divergence
p	frequency counts of distributions
q	frequency counts of distributions
x	the unique discrete x values of P and Q: x = unique([P;Q])

Note

reimplemented from matlab

Author(s)

Michael Thrun

Mthrun/Distances documentation built on Feb. 4, 2020, 8:39 p.m.