trainnet.m

一个关于adaboost和NN的matlab程序

function [net,epochs,rms] = trainnet(in,targets,S1,validation)
% WBL 22 August 2002 Use P450 training data to train Matlab MLP Nural Network
% $Revision: 1.5 $  $Date: 2002/10/31 11:22:31 $

%WBL 13 Sep 2002 as trainnet - but pass net via arguments, rather than saving to file
%WBL  9 Sep 2002 implement early stopping
%WBL  9 Sep 2002 Based on read.m 

%    DEFINING THE NETWORK
%    ====================
% S1 = number of nodes in hidden layer
% S2 = number of output nodes (ie 2)
S2 =  2;

net = newff(minmax(in),[S1 S2],{'logsig' 'logsig'},'traingdx');
%net.LW{2,1} = net.LW{2,1}*0.01;
%net.b{2} = net.b{2}*0.01;

%    TRAINING THE NETWORK WITHOUT NOISE
%    ==================================

net.performFcn = 'sse';        % Sum-Squared Error performance function
%net.trainParam.goal = 0.1;     % Sum-squared error goal.
net.trainParam.show = inf;      % Frequency of progress displays (in epochs).
net.trainParam.epochs = 5000;  % Maximum number of epochs to train.
%net.trainParam.epochs = 5;  % Maximum number of epochs to train.
%net.trainParam.mc = 0.95;      % Momentum constant.

%    Training begins...please wait...

%[net,tr] = train(net,in,targets);
[net,tr] = train(net,in,targets,[],[],validation);
%tr
epochs = tr.epoch(size(tr.epoch,2));
rms = sqrt(tr.perf(size(tr.perf,2))/size(in,2));
%filename = sprintf('13-sep-2002_%dx%d', S1, ntests);
%save (filename, 'net');