Comparison of input data compression methods in neural network solution of inverse problem in laser Raman spectroscopy of natural waters

Abstract

In their previous papers, the authors of this study have suggested and realized a method of simultaneous determination of temperature and salinity of seawater using laser Raman spectroscopy, with the help of neural networks. Later, the method has been improved for determination of temperature and salinity of natural water using Raman spectra, in presence of fluorescence of dissolved organic matter as dispersant pedestal under Raman valence band. In this study, the method has been further improved by compression of input data. This paper presents comparison of various input data compression methods using feature selection and feature extraction and their effect on the error of determination of temperature and salinity. © 2012 Springer-Verlag.