MODEL-BUILDING WITH NEURAL NETWORKS BASED ON GENETIC ALGORITHMS FOR STRUCTURE-ACTIVITY-RELATIONSHIPS OF PESTICIDE

In this paper, a neural network based on genetic algorithms was propos ed for a QSAR model building of herbicidal N-(1-methyl-1-phenylethyl) phenylacetamides. Since neural networks can express complicated struct ure-activity relationships, and the combination of Genetic Algorithms can help the networks to jump out of the local optimal points, thus, i t speeds up the covergence of the training. At the same time, it's pre diction precision has been notably raised.