mmvn_toolkit

complete toolkit for generating, training, testing and viewing multidimensional gaussian mixtures

Description

A toolkit for Gaussian mixtures.
flexible tools for:
Generating univariate, multivariate, or mixtures of gaussians
Interactive viewing tools allows viewing of multidimensional data and models. Initialize models, add and remove dimensions or clusters and inspect the fit in real-time.
Also includes tools to subset the data using model-based (pseudo-)metrics.

simple example of clustering
% generate 4 clusters in 2 dimensions
  [X idx theta] = mmvn_gen( 1000, [0 5; 5 0; 5 5; 0 0] );

% fit data
  Opt = mmvn_fit( X, 4, theta );

% plot data
  scatter( X(:,1), X(:,2), [], idx, 'filled');
  hold on;
% represent fits using ellipses
  ellipse( Opt.M, Opt.V);
 

See mmvn_tutorial (included) for a walk-through of major functionality

Acknowledgements

The author wishes to acknowledge the following in the creation of this submission:
PSNR

• MATLAB release: MATLAB 7.4 (R2007a)

[foued saadaoui]

 

[Shawn_lj Qingsheng]

Hello! I tried the example in the MMView.m, but the error happened as below:
>>mm = MModel(X, [1 2], 4 );
??? Error using ==> kmeans.kmeans
Too many input arguments.

Error in ==> MModel.kmeans_init at 30
    [Ci,C] = kmeans(mm.x(:,mm.s), k,...

Error in ==> MModel.MModel at 90
        mm = kmeans_init(mm,t);

Please solve the problem and tell me, thanks!

[vaishali kumar]

toolkit for gaussian mixture model in audio
 

[vaishali kumar]

hello,any toolkit for gaussian mixture model based audio source separation