神经网络时间序列应用程序处理超出数据范围的值

Question

我想自定义我的神经网络训练以在[0,65]的范围内进行处理 - 我尝试了代码：

net.inputs{1}.processParams;
ymin = 0;
ymax = 65;

net.outputs{2}.processParams;
ymin = 0;
ymax = 65;

然而，当我检查时间序列 - 响应图时，它仍会处理超出这两个值的值。

使用：

SampleData:
18
39
51
26
13
9
13
 9
 13
 13
 13
 4 
 46 
 52
 52
 8

T = tonndata(SampleData,false,false);
trainFcn = 'trainbr';
feedbackDelays = 1:2;
hiddenLayerSize = 400;
net = narnet(feedbackDelays,hiddenLayerSize, 'open', trainFcn);
[x,xi,ai,li] = preparets(net,{},{},T);

net.divideParam.trainRatio = 70/100;
net.divideParam.valRatio = 15/100;
net.divideParam.testRatio = 15/100;

然后......扩大训练处理......

 net.inputs{1}.processParams;
ymin = 0;
ymax = 65;

net.outputs{2}.processParams;
ymin = 0;
ymax = 65;