A comparison of backpropagation and generalized-regression genetic-neural network models.

Abstract

The results of the backpropagation (BP) and generalized-regression genetic-neural (GRGN) network were compared using a series of nonpeptide arginine vasopressin VI antagonists. It was shown that both approaches are equivalent with respect to the recognition process while the BP network is superior over GRGN if the sample sizes are lowered by cross-validation.