MATLAB Answers

An error occuered before neural network training

1 ビュー (過去 30 日間)
kengo fujimura
kengo fujimura 2019 年 7 月 12 日
回答済み: Hiroyuki Hishida 2019 年 10 月 16 日
Hi
I want to classify time series data by unsupervised learning
So,I'm making recurrent self organized neural network by Shallow Neural Network.
This error occuered
error: +
The dimensions of the matrix must match.(Original txt:行列の次元は一致しなければなりません。)
error: trainbu>train_network (line 196)
net.IW{i,j} = net.IW{i,j} + dw;
error: trainbu (line 52)
[out1,out2] = train_network(varargin{2:end});
error: network/train (line 353)
[net,tr] = feval(net.trainFcn,'apply',net,tr,data,calcMode.options,hints,net.trainParam);
error: SOM_test (line 79)
[net2,tr] = train(net2,Xs,{},Xi,Ai);
Program I wrote is this
[X, Y]= simpleseries_dataset;
net2.userdata='fujimura_kengo';
dimension1 = 5;
dimension2 = 5;
net2= layrecnet(1:2,10);
net2.numLayers=2;
net2.name=('Den-O');
net2.trainFcn='trainbu';
net2.numInputs=1;
net2.layers{1}.initFcn='initwb';
net2.layers{1}.netInputFcn='netsum';
%2層目の設定
net2.layers{2}.dimensions=[5 5];
net2.layers{2}.distanceFcn='linkdist' ;
net2.layers{2}.initFcn= 'initwb' ;
net2.layers{2}.netInputFcn= 'netsum' ;
net2.layers{2}.topologyFcn= 'hextop' ;
net2.layers{2}.transferFcn= 'compet' ;
net2.layers{2}.netInputParam='netprod';
net2.layers{2}.transferParam='elliotsig';
net2.layers{2}.distanceParam='dist';
net2.inputConnect=[1 ; 0 ];
net2.layerConnect=[1 0; 1 0 ];
net2.outputConnect=[0 1 ];
net2.biasConnect=[0;0];
net2.divideFcn=''
view(net2);
[net2,tr] = train(net2,Xs,{},Xi,Ai);
I want fix this error.But I can't ditect cause of error
If someone know how to fix,please teach me how to fix

回答 (1 件)

Hiroyuki Hishida
Hiroyuki Hishida 2019 年 10 月 16 日
Hi Kengo,
I tried reproduce your phenomemon, but it was failed since 'Xs', 'Xi' and 'Ai' are not defined.
Could you add more information?
Best
Hiroyuki

製品


リリース

R2017a

Community Treasure Hunt

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

Start Hunting!