ANN Problem: matrix dimensions must agree.

Question

Sitra 2017-6-7

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/343743-ann-problem-matrix-dimensions-must-agree

回答： Greg Heath 2017-6-7

Hi Dears,

I'm facing the following error message:

Matrix dimensions must agree.
Error in nn7.grad2 (line 104)
    gN{i} = Fdot .* gA{i};
Error in nn7.perfs_sig_grad (line 23)
  [gB,gIW,gLW,gA] = nn7.grad2(net,trainData.Pc,trainData.Pd,...
Error in trainb>train_network (line 149)
    nn7.perfs_sig_grad(net,trainData,valData,testData,needGradient,fcns);
Error in trainb (line 56)
            [out1,out2] = train_network(varargin{2:end});
Error in network/train (line 353)
    [net,tr] = feval(net.trainFcn,'apply',net,tr,data,calcMode.options,hints,net.trainParam);
Error in ANN2 (line 90)
[net,tr] = train(net,Xs,T);

I tried to figure-out the sources of this problem, and I get the following indication:

when I put the following command just before train command, I get:

whos net
[net,tr] = train(net,Xs,T);
Name       Size               Bytes  Class      Attributes
T         18x1000            144000  double               
Xs        18x1000            144000  double               
net        1x1                44008  network

Every time I run the program, the preceding error pops-up when net's bytes limit decreases down to 44008!!!

It seems to me that there is a minimum limit of network's bytes, because it happens when net's byte is less than 44008 (or it could happen when it is slightly larger than 44008)

What is the exact minimum information needed for the network, and how can we avoid the above problem?

I have checked many questions raised in this forum and no one works with me!

Thank you so much

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Sitra 2017-6-7

New observations that may help others:

I found that error when I use some training functions, such as: trainb, traingdx, trains, traingdm, traincgf

Should I normalize my data for such these training functions? because I heard that MATLAB's fitnet and other functions automatically normalize and de-normalize the dataset?

Thank you again

请先登录，再进行评论。

请先登录，再回答此问题。