Pattern Recognition Approach to Vapor Phase Infrared Spectra Interpretation for Gas Chromatography

Abstract
The ability to distinguish functional group and the presence of specific elements in vapor phase infrared spectra is demonstrated using pattern recognition techniques. Kth nearest neighbors analysis was found to be superior to hyperplane separations. Weight-sign feature selection was used to elucidate the significant wavelengths which exhibited agreement with established group frequencies. The use of vapor phase infrared spectra for gas chromatographic peak identification is discussed.