ELUCIDATION OF CHEMICAL-REACTIVITY USING AN ASSOCIATIVE MEMORY SYSTEM

An associative memory system, a model of the information processing in the human cerebellum, is used to investigate chemical reactivity in terms of polar bond breaking. The information which bonds in an organic molecule break or do not break, an information discernible from reaction databases, is generalized. An optimization procedure was introduced to find the associative memory system with the highest predictive power. In this way, the electronic and energy effects that have the greatest influence on chemical reactivity can be determined. The results are compared with those obtained from various pattern recognition methods. An associative memory system is able to represent the relationship between structure and reactivity in an implicit manner without the user having to specify an explicit mathematical relationship. Once trained it can be successfully applied to the prediction of chemical reactivity for a wide range of organic chemistry. This is demonstrated with some examples.