基于改进的遗传算法的特征选择方法在冠心病检测中的应用

Application of a Feature-Selection Method Based on Improved Genetic Algorithm in Coronary Heart Disease Detection

提出了一种基于改进的遗传算法的特征选择方法(MGA,Modified Genetic Algorithm),并结合支持向量机分类模型在冠心病数据库上进行分类.算法采用基于特征的F-Score值的先验知识指导种群初始化,并在每次迭代过程中生成同样规模的子代,与父代竞争,进而扩大解的搜索空间,使种群个体种类更加丰富,以改进早熟收敛问题.当遗传算法进化速度变慢时,采用类模拟退火的方式对遗传算法进行刺激,使遗传算法尽可能地跳出局部最优解的束缚,从而趋向全局最优解.实验结果表明,该算法有效地对特征进行降维,并提高了算法的分类性能.

This paper proposes a feature-selection method based on an improved genetic algorithm,combined with the support vector machine classification model to classify the coronary heart disease.In order to alleviate the problem of premature convergence,the algorithm uses the prior knowledge of F-Score to guide the population initialization,and generates offspring with the same size of the parent during each iteration,which competewith the parent.This measure expands the search space of the solution,and makes the species of the population more abundant.When the evolution speed of the genetic algorithm slows down,a similar simulated annealing method stimulates the genetic algorithm to jump out of the constraints of the local optimal solution as much as possible,so as to approach the global optimal solution.Experimental results show that the algorithm effectively reduces the dimensionality of features and improves the classification performance of the algorithm.