lyngby_nn_emain - Main function for Entropic Neural network
function [V, W] = lyngby_nn_emain(X, T, PropertyName, ...
PropertyValue)
Input: X Neural network input
T Target output
Property: GenOptim { {Free} | EarlyStop |
HiddenUnitsEarlyStop | Pruning
Pruning1DRegGridSearch |
Pruning2DRegGridSearch ]
Generalization optimization
Validation [ {SingleBlocked |
BlockedCrossvalidation ]
How to compute the generalization
measure
TrainSet [ {not defined} | <indices> ]
Indices for the training
set. Default is all of X.
ValSet [ {not defined} | <indices> ]
Indices for the validation
set. Only used in connection with
"non-free" generalization
optimization. Default is no
validation set.
HiddenUnits { 3 } Number of hidden units, not
counting the threshold unit
Reg { 0.001 } Regularization parameter
(weight decay)
Seed Seed for the random generator for
the weights Matlab 4.x single seed
generator is used. Default is no
seed.
Info [ {0} | 1 ] Continuous information
about the optimization
Output: V Input weights
W Output weights
EAcc Evolution of error. Is dependent on the setting
of 'GenOptim'
Info Information about EAcc
Main function for the "entropic neural network" that is the
neural network with one binary output.
See also LYNGBY, LYNGBY_NN_QMAIN, LYNGBY_NN_CMAIN,
LYNGBY_NN_ETRAIN.
$Id: lyngby_nn_emain.m,v 1.24 2002/03/18 18:18:52 fnielsen Exp $