LSTM training problem in MATLAB

Question

Hossein Malekahmadi 2021-7-24

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/885214-lstm-training-problem-in-matlab

评论： Sruthi Gundeti 2021-7-26

Hi i tried to run LSTM with below code and i dont know why i get this error

close 
clc
% Calculating amount of test data
N = round(size(inp_train,1)*30/100);%change 0.3 for defferent amount of test data
% Seperating data for training and testing
nn_train = inp_train(1:end-N,:);
in_test = inp_train(end+1-N:end,:);
tn_test = tar_train(end+1-N:end,:);
nn_target = tar_train(1:end-N,:);
numfeatures = size(nn_train,2);
numHiddenUnits = 100;
numClasses = size(nn_target,2);
layers = [...
    sequenceInputLayer(numfeatures)
    lstmLayer(numHiddenUnits)
    fullyConnectedLayer(numClasses)
    regressionLayer];
maxEpochs = 1000;
options = trainingOptions('adam',...
    'MaxEpochs',maxEpochs,...
    'InitialLearnRate',0.0001,...
    'GradientThreshold', 0.01);
net = trainNetwork(nn_train, nn_target, layers, options);

the error:

Error using trainNetwork (line 150)

Invalid training data. Sequence responses must have the same sequence length as the corresponding

predictors.

Error in LSTM (line 30)

net = trainNetwork(nn_train, nn_target, layers, options);

and my data:

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Sruthi Gundeti 2021-7-26

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/885214-lstm-training-problem-in-matlab#answer_753729

Hi,

The LSTM network considers inputs as follows

No of rows= No of features

No of columns= No of Samples

No of samples for training data and target data must be same i.e., No of columns of NN_target and nn_train must be same

You can train by using transpose of both data

nn_train=nn_train'

nn_target=nn_target'

i.e.,net = trainNetwork(nn_train, nn_target, layers, options);

2 个评论
显示无隐藏无

Hossein Malekahmadi 2021-7-26

编辑：Hossein Malekahmadi 2021-7-26

Thank you for your respond to my question when I checked the codes you mentioned I realised that I forget to transposing my data Again thank you very much Sruthi Gundeti

Sruthi Gundeti 2021-7-26