Classification and prediction of reagents' roles by FRAU system with self-organizing neural network model

The classification and prediction of the roles for reagents in reactions ar e presented. The same dimensional representation of various reagents indepe ndent of the number of atoms was achieved by selecting representative facto rs by the FRAU (Field-characterization for Reaction Analysis and Understand ing) system. Training of a self-organizing model considering both negative and absent data was accomplished by modifying the original counter-propagat ion (CP) type of Kohonen neural network to treat absent data differently fr om negative data. The modified CP Kohonen neural network successfully class ified the reagents and produced a reagent-roles correlation model that give s good answers predicting roles of the reagents.