Hello
MATLAB’s feedforward neural networks expect the input data as a matrix of size [numFeatures × numSamples] and the output (target) data as [numOutputs × numSamples].
For this scenario, the input data should be arranged as a matrix with 5 rows and 120 columns (representing 5 features for each of the 120 samples), and the output data should be a matrix with 2 rows and 120 columns (representing 2 outputs per sample), rather than using cell arrays.
An approach to resolve the issue is to reshape the data into matrices and use the “feedforwardnet” function for simplicity. Refer to the sample code snippet below:
XTrainMat = cell2mat(XTrain'); % 5×120
YTrainMat = cell2mat(YTrain'); % 2x120
net = feedforwardnet(4);
net.layers{1}.transferFcn = 'logsig';
net = train(net, XTrainMat, YTrainMat);
view(net);
To know more about the “feedforwardnet” function, visit the link below:
Hope it helps
