GANDERS |
Solves the Generalized Anderson's task.
Synopsis:
model = ganders( distrib)
model = ganders( distrib, options)
model = ganders( distrib, options, init_model )
Description:
This function is an implementation of the general framework
to find the optimal solution of the Generalized Anderson's
task [SH10].
The goal of the GAT is find the binary linear classification
rule (g(x)=sgn(W'*x+b) with minimal probability of
misclassification. The conditional probabilities are known to
be Gaussians their paramaters belong to a given set of parameters.
The true parameters are not known. The linear rule which
guarantes the minimimal classification error for the worst
possible case (the worst configuration of Gaussains) is
sought for.
Input:
distrib [struct] Set of binary labeled Gaussians.
.Mean [dim x ncomp] Mean vectors.
.Cov [dim x dim x ncomp] Covariance matrices.
.y [1 x ncomp] Labels of the Gaussians (1 or 2).
options [struct] Determines stopping conditions:
.tmax [1x1] Maximal number of iterations (default inf).
.eps [1x1] Minimal improvement of the optimized
criterion (default 1e-6).
.mineps_tmax [1x1] Number of iterations of the one-dimensional
numerical search (default 100).
init_model [struct] Initial model:
.W, .b, .t see below.
Output:
model [struct] Binary linear classifer:
.W [dim x 1] Normal vector of the found hyperplane W'*x + b = 0.
.b [1x1] Bias of the hyperplane.
.r [1x1] Mahalanobis distance for the cloasest Gaussian.
.err [1x1] Probability of misclassification.
.t [1x1] Number of iterations.
.exitflag [1x1] 0 ... maximal number of iterations was exceeded.
1 ... solution was found.
-1 ... solution (with err < 0.5) does not exist.
Example:
distrib = load('mars');
model = ganders( distrib );
figure; pandr( model, distrib );
See also
ANDRORIG, EANDERS, GGRADANDR, ANDRERR, LINCLASS.
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:
4-may-2004, VF
17-sep-2003, VF