Why the neural network creates the same output values for the different inputs?

2 次查看（过去 30 天）
Perihan Bilgeç
Perihan Bilgeç 2018-11-3
评论： Perihan Bilgeç 2018-11-7
I created a neural network model as you can see the model code. Firstly I normalized the data between 0 and 1, then I divided the input and target data for test and validation. I see that the all validation output values (y) is the same. I checked the normalized validation input data, it doesn't contain an error.
(Xtn: Normalized input test values, Xvn: Normalized input validation values, Ytn: Normalized target test values, Y:Normalized validation output values, Ye: Denormalized validation output values)
Can you help me?
Data=xlsread('Data31.10.xlsx'); Input=Data(:,4:12); Target=Data(:,end);
I=Input';
T=Target';
A=minmax(I); Min=A(:,1); Max=A(:,2); Fark=Max-Min; In=(I-Min)./Fark;
Tmin=min(T); Tmax=max(T);
nofdata=size(In,2);
ntd=round(nofdata*trainingrate);
Xtn=In(:,1:ntd);
Xvn=In(:, ntd+1:end);
Yt=T(1:ntd);
Yv=T(ntd+1:end);
Ytn=(Yt-Tmin)./(Tmax-Tmin);
net=newff(Xtn, Ytn, [n1,n2], {'logsig', 'logsig', 'logsig'}, 'trainlm');
net=train(net,Xtn,Ytn);
y=sim(net,Xvn);
ye=y*(Tmax-Tmin)+Tmin;
  3 个评论
显示 1更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

回答（1 个）

Greg Heath
Greg Heath 2018-11-7
Where is your training data?
Typically, the data division is
train/val/test = 70%/15%/15%
Hope this helps,
Greg

类别

在 Help Center 和 File Exchange 中查找有关 Define Shallow Neural Network Architectures 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by