I have 3 different wavelength signals to use as an input to predict single output. BTW I have extracted 9 features per each signals. Now I'm training the classification NN with Matlab. I want to use 30 datasets (each set contain of 3 different wavelength) of signals to classify into 3 classes. what should my input matrix be? I have tried to create a cell of 3 different signals inside but the nprtool is not working with it. Please help me, I am new and not familiar with Neural Network.
1 Answers
0
votes
I assumed signal is vector and output will be the class value in binary so we need 3 digits for 9 classes.
x = [signl1 signl2 singl3]; %all verctors in one row
label = [0 1 0]; %class 2 in binary
net = patternnet(10);
view(net)
[net,tr] = train(net,x,label);
Now you can use net
as signal classifier for new signals after validating your NN with dataset you never use before.
estimiated_class = net(new_signal)
this is my vague idea if not work try deep learning