Classification of Raman Spectra using Support Vector Machines

The classification of Raman Spectra is useful in identification and diagnosis applications. We have obtained Raman Spectra from bacterial samples using three different species of bacteria. Before any form of classification can be carried out on the Raman Spectra it is important that some form of normalization is used. This is due to the nature of the readings obtained by the acquisition equipment. The method used for normalization greatly affects the accuracy of the results. We have carried out experiments using Support Vector Machines and the correlation kernel. Our observations have led us to the hypothesis that the correlation kernel is "self-normalizing" and gives satisfactory results without the need of any other normalization technique.