Help with gmdistribution and feeding model into function

Hi Team!

I have a question and I hope that someone can perhaps show me where I am missing something. I am trying to implement the Hellinger Distance. The Matlab code can be found here. Try as I might, I cannot get it to work. I wrote the toy program below to try to figure out the issue and I still can't get it working. My code is:

clear all
clc
 
mu1 = [1 2;-3 -5];
sigma1 = [1 1]; % shared diagonal covariance matrix
 
mu2 = [1 4;-1 4];
sigma2 = [1 1]; % shared diagonal covariance matrix
 
gm1 = gmdistribution(mu1,sigma1)
gm2 = gmdistribution(mu2,sigma2)
 
% Calculate Hellinger Distance
% Error in this single line:
H = uHellingerJointSupport2_ND(gm1,gm2)

Now I should be getting a single number between 0 and 1 but I am getting the following errors:

Unrecognized method, property, or field 'Mu' for class 'gmdistribution'.

Error in classreg.learning.internal.DisallowVectorOps/subsref (line 22)

[varargout{1:nargout}] = builtin('subsref',this,s);

Error in mergeDistributions (line 16)

model.Mu = [ model1.Mu, model2.Mu ] ;

Error in uHellingerJointSupport2_ND>main_uHellinger (line 77)

f0 = mergeDistributions( f1, f2, [0.5 0.5], 0 ) ;

Error in uHellingerJointSupport2_ND (line 51)

H = main_uHellinger(f1, f2, useWeightedHellinger) ;

I have gone through the individual functions and I just don't understand why my toy example will not work. If anyone can provide some insights, I would be very thankful.

Answer 1

[Cris LaPierre]

I think the issue is you are trying to use the MATLAB function gmdistribution to create the input for the file exchange function uHellingerJointSupport2_ND. However, gmdistribution returns a gmdistribution object, while uHellingerJointSupport2_ND expects the inputs to be structure arrays with the following fields:

• Mu
• w
• Cov

[dsmalenb]

Oh I see! I will need to try that later today. Thanks for the thought! I will post how that works out.