I have a 1D CNN for the regression task. The input of my CNN is a 1D signal and the parameters included in this signal are: amplitude, damping and frequency shift; let's say a,d and f:
Signal(t) = f(a,d,f)
I trained my network just with amplitude; which its result is not bad at all; however, I was wondering to know it is possible to include other parameters in output?! in this way the output of the regression layer must have 3 channels? my framework is Matlab deep learning toolbox.
Anouar Ben Khalifa Thank you very much. What is the point of using LSTM. I have not used LSTM for the regression task, I appreciate it if get an example or paper. I am a little bit worried because my signal is not the usual time series signal.
Dear Sir
this example on MATLAB can help you.
Good luck
https://www.mathworks.com/help/deeplearning/ug/sequence-to-sequence-regression-using-deep-learning.html
the output depend on target, if you enter many inputs and one target, so the output must be one.
- Badges
- Science topic