Training LTSM network error in command window

3 次查看（过去 30 天）
G.N. Cobbold
G.N. Cobbold 2020-11-28
回答： Rishik Ramena 2020-12-7
'Invalid training data. Responses must be a vector of categorical responses, or a cell array of categorical response
sequences.'
Hello I am currently training a LTSM classifaction network, in which i keep coming into this same error line. I have two data types. Firstly, my xtrain is a cell array of (3 x 900 double) whilst my ytrain is a cell array with categorical vectors( 3x 900 categorical). I am running this code to provide me a netowrk which will window temproal data of the sensor readings. I initally suspected that the error was due to my ytrain, in which i have countlessly spent time to change this data type to get desired result. Could this error be po potentially an error with my xtrain?
[net] = trainNetwork(xtrain,ytrain,layers,options);
numFeatures = 1;
numHiddenUnits = 200;
numClasses = 3;
trainFcn = 'trainscg'; % Scaled conjugate gradient backpropagation.
layers = [ ...
sequenceInputLayer(numFeatures)
lstmLayer(numHiddenUnits,'OutputMode','sequence')
fullyConnectedLayer(numClasses)
softmaxLayer
classificationLayer];
options = trainingOptions('adam', ...
'MaxEpochs',250, ...
'GradientThreshold',1, ...
'InitialLearnRate',0.005, ...
'LearnRateSchedule','piecewise', ...
'LearnRateDropPeriod',125, ...
'LearnRateDropFactor',0.2, ...
'Verbose',0, ...
'Plots','training-progress');
%'MaxEpochs',60, ...
%'GradientThreshold',2, ...
%'Verbose',0, ...
%'Plots','training-progress');
%ytrain=categorical(ytrain);
%ytrain = transform(ytrain,@(data) padSequence(data,sequenceLength));
[net] = trainNetwork(xtrain,ytrain,layers,options);
%net = trainNetwork(xtrain,Ytrain,layers,options);
t_training=t_training+toc;

请先登录，再回答此问题。

回答（1 个）

Rishik Ramena
Rishik Ramena 2020-12-7
You probably need to convert the ytrain to categorical while passing to trainNetwork.
ytrain=categorical(ytrain);
[net] = trainNetwork(xtrain,ytrain,layers,options);

类别

在 Help Center 和 File Exchange 中查找有关 Sequence and Numeric Feature Data Workflows 的更多信息

标签

产品


版本

R2020b

Community Treasure Hunt

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

Start Hunting!

Translated by