Generate data drawn form a mixture of Gaussians.
For definitions of separation and eccentricity see:
Sanjoy Dasgupta, "Learning Mixtures of Gaussians", FOCS, 1999.
http://cseweb.ucsd.edu/~dasgupta/papers/mog.pdf
USAGE
[X0,H0,X1,H1] = demoGenData(n0,n1,k,d,sep,ecc,[frc])
INPUTS
n0 - size of training set
n1 - size of testing set
k - number of mixture components
d - data dimension
sep - minimum separation degree between clusters (sep > 0)
ecc - maximum eccentricity of clusters (0 < ecc < 1)
frc - [0] frac of points that are noise (uniformly distributed)
OUTPUTS
X0 - [n0xd] training set data vectors
H0 - [n0x1] cluster membership in [1,k] (and -1 for noise)
X1 - [n1xd] testing set data vectors
H1 - [n1x1] cluster membership in [1,k] (and -1 for noise)
EXAMPLE
n0=1000; k=5; d=2; sep=2; ecc=1; frc=0;
[X0,H0,X1,H1] = demoGenData(n0,n0,k,d,sep,ecc,frc);
figure(1); clf; visualizeData( X0, 2, H0 ); title('train');
figure(2); clf; visualizeData( X1, 2, H1 ); title('test');
See also visualizeData, demoCluster
Piotr's Computer Vision Matlab Toolbox Version 3.20
Copyright 2014 Piotr Dollar. [pdollar-at-gmail.com]
Licensed under the Simplified BSD License [see external/bsd.txt]