| MLSIGMOID | |
Fitting a sigmoid function using ML estimation.
Synopsis:
model = mlsigmoid(data,options)
Description:
model = mlsigmoid(data,options) computes Maximum-Likelihood
estimation of parameters of sigmoid function [Platt99a]
p(y==1|x) = 1/(1+exp(A(1)*x+A(2))),
used to describe a posteriory probability of a hidden binary
state y from {1,2}. The conditional probabilities p(x|y) are
assumed to be uni-variate Gaussian distribution. The training
samples {(X(1),y(1)),...,(X(num_data),y(num_data))} assumed to
be i.i.d. are given in data.X and data.y.
Input:
data [struct] Input sample:
.X [1 x num_data] Values of discriminant function.
.y [1 x num_data] Corresponding class label (1 or 2).
options [struct] Control parameters:
.regul [1x1] If 1 then fitting is regularized to prevent
overfitting (default 1).
.verb [1x1] If 1 then progress info is displayed (default 0).
Output:
model.A [1x2] Parameters of sigmoid function.
model.logl [1x1] Value of the log-likelihood criterion.
Example:
help demo_svmpout;
See also
SIGMOID.
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:
28-apr-2008, VF; fixed incorrect regularization for the positive labels
03-jun-2004, VF
11-oct-2003, VF
20-sep-2003, VF
08-may-2003, VF