Improving signal to noise ratio of Raman spectra is vital for the application. Here, authors show a noise learning method that learns the noise feature of a spectrometer. This improves the signal to noise ratio and makes deep learning to be instrument dependent instead of sample dependent.
