一种基于BP模型的分子致癌性评价方法

EVALUATION OF MOLECULAR CARCINOGENICITY BY BACK PROPGATION NEURAL NETWORKS

化学物质的致癌性监测技术的研究已成为国际性科研热点，采用分子构效关系对分子致癌性进行评价是一种区别于生物学检测的一种新方法。本文介绍一种以分子拓扑结构为参数、用反向传播神经网络ＢＰ模型建立的分子致癌性评价方法，通过一组多环芳烃分子进行训练和测试，讨论ＢＰ模型在此应用中的有关问题。

Devoloping biological methods for monitoring carcinogens re in world wide hot point,meanwhile a new method appears which is based on molecular Stucture Activity Relationship(SAR).This paper describes a method which uses BP neural network as evalation model,in which molecular topological structure is the structure parameter,and discusses the application of BP model through the trainning and testing of a group of multicyclic aromatic chemicals.