JOINT NEURAL-NETWORK INTERPRETATION OF INFRARED AND MASS-SPECTRA

Combining gas phase infrared (IR) spectra with mass spectral (MS) data , a neural network has been developed to predict 26 different molecula r substructures from multispectral information. The back-propagation p rocedure has been used for training, including its previously publishe d modification, the flashcard algorithm. Present functional groups hav e been detected correctly in 86.4% of all cases, compared with 88.4% u sing only IR and 78.2% using only MS data for training and prediction. For only 8 out of the 26 functionalities does the joint utilization o f infrared and mass spectra yield better prediction results, with the greatest improvement being for halogen bond predictions. The predictio n of functional group absence results in accuracy of about 95.5% for b oth IR and IR/MS networks but only 87.1% for a stand alone MS network. Insights have been gained into the suitability of both data sets for neural network training by presenting just IR or MS data to a jointly trained neural network, revealing the amount of information the networ k utilizes from either spectroscopic technique. In addition, an algori thm which produces balanced training and test sets for multi-output ne ural networks has been devised.