Abstract

A method for identifying bacterias was studied by combining Uv-vis transmission spectra with support vector machine. 3 fold cross validation based on grid search method was used to obtain the best penalty factor C and the kernel parameter, the rapid identification model was built by One-against-one classification method. The model could identify the four species of bacterias as the accuracy was 100%, and different subspecies of E. coli could also be identified. The identification model had good accuracy, stability and generalization capability.

© 2015 Optical Society of America