CONNECTIONIST HYPERPRISM NEURAL-NETWORK FOR THE ANALYSIS OF ION MOBILITY SPECTRA - AN EMPIRICAL-EVALUATION

Spectra from ion mobility spectrometry (IMS) consist of low resolution information arising from complex gas-phase ion-molecular reactions. S uch spectra lack specific structural information and have been conside red inappropriate for analysis by traditional library search and patte rn recognition methods. A data set of 292 mobility spectra from 12 org anic compounds was treated using a connectionist hyperprism classifica tion (CHC) neural network. Advantages of the CHC network for such spec tra included high speed, linear training, and flexibility to adapt tra ining parameters, to minimize false negative identifications. The prin cipal disadvantages were a sensitivity to unnecessary spectral informa tion in the training vectors and a high false positive ratio. However, sensitivity exhibited by the CHC neural network assisted in identifyi ng dispensable spectral information and in suggesting the importance o f spectral features.