RSDE | ![]() |
Reduced Set Density Estimator.
Synopsis:
model = rsde(X,options)
Description:
This function implements the Reduced Set Density Estimator
[Girol03] which provides kernel density estimate optimal
in the L2 sense. The density is modeled as the weighted sum
of Gaussians (RBF kernel) centered in seleceted subset of
training data.
Input:
X [dim x num_data] Input data sample.
options [struct] Control parameters:
.arg [1x1] Standard deviation of the Gaussian kernel.
.solver [string] Quadratic Programming solver (default 'quadprog');
Output:
model [struct] Output density model:
.Alpha [nsv x 1] Weights of the kernel functions.
.sv.X [dim x nsv] Selected centers of kernel functions.
.nsv [1x1] Number of selected centers.
.options.arg = options.arg.
.options.ker = 'rbf'
Example:
gnd = struct('Mean',[-2 3],'Cov',[1 0.5],'Prior',[0.4 0.6]);
sample = gmmsamp( gnd, 200 );
figure; hold on; ppatterns(sample.X);
plot([-4:0.1:8], pdfgmm([-4:0.1:8],gnd),'r');
model = rsde(sample.X,struct('arg',0.7));
x = linspace(-4,8,100);
plot(x,kernelproj(x,model),'g');
ppatterns(model.sv.X,'ob',13);
See also
KERNELPROJ.
About: Statistical Pattern Recognition Toolbox
(C) 1999-2003, Written by Vojtech Franc and Vaclav Hlavac
Czech Technical University Prague
Faculty of Electrical Engineering
Center for Machine Perception
Modifications:
17-sep-2004, VF, revised