基于遗传算法的非线性映射预测硝基苯化合物的生物毒性  

Nonlinear mappimg based on genetic algorithms to predict biological toxicity of nitrobenzene derivatives

作　　者：李仲[1] 王凤英 

机构地区：[1]青海民族大学电子工程与信息科学系,西宁810007

出　　处：《原子与分子物理学报》2010年第5期824-828,共5页Journal of Atomic and Molecular Physics

基　　金：国家民委科研基金(08QH02)

摘　　要：非线性映射是一种降维处理技术.通过Sammon算法可以将多维分析问题转化为二维问题分析,应用该法的关键在于寻求二维空间的最佳投影点,这可以转化为一个复杂的非线性优化问题来进行解决,遗传算法是解决各类函数优化和非线性映射问题的一种有效算法.以硝基苯化合物的生物毒性的模式识别为例,编制基于误差函数的目标函数和遗传算法约束参数,应用基于MATLAB的遗传算法工具箱进行优化求解.结果表明:该工具箱在求解此类非线性优化问题上是有效的,基于遗传算法的非线性映射的预测结果比其它方法要好些.Nonlinearity mappimg is a dimension--reduced data treatmenf method. It can change many dimensions＇ question into two dimension question by Sammon algorithm. It is difficult to search the best projection point of two dimension, but it can be solved by transforming a nonlinear optimization question. Genetic Algorithms are effective algorithms for solving all kinds of optimization problems and nonlinearity mappimg. By compiling object function based on error function and constraint parameters, a model of projection pursuit on pattern recognition of biological toxicity of nitrobenzene derivatives was solved based on based on MATLAB genetic algorithm Toolbox. Result indicated that the toolbox was efficient. It has been found that the prediction result by nonlinear mappimg based on genetic algorithms is better than that of other method.

关 键 词：遗传算法 非线性映射 硝基苯化合物 生物毒性 MATLAB 

分 类 号：O658[理学—分析化学]