Study on Computer-assisted Infrared Spectroscopy for Identification of Chemical Structure System

Abstract

In the past decades, people are trying to search the way to analyze the infrared spectra. Along with the computerization of the commercialized infrared spectroscopy, there are many computer-assisted identification methods of infrared spectroscopy. For decades, people have been exploring the empirical analysis of infrared spectroscopy. These methods can be divided into three categories: expert system; spectrum retrieval system and pattern recognition method. The most commonly used pattern identification methods are artificial neural network and partial least squares. The literature shows that the prediction accuracy of the structural fragments is not very high, and the neural network is still unstable, easy to fall into the local optimal and slow convergence and other issues. In this paper, the support vector machine is used to analyze the sub-structure of infrared spectroscopy. The vector machine is a good machine learning algorithm for small sample system. For most of the substructures, the predictive ability of support vector machines is better. The support vector machine also has the advantages of stability and fast training speed. It is a good tool for assistant analysis of infrared spectrum.