改进GA-SVM在冠状动脉疾病诊断中的应用

改进的遗传算法（GA）自动优化支持向量机（SVM）参数,同步决策最优特征子集。新颖的分组多基因交叉技术保留了基因小组中的信息,而且允许后代继承更多的来自染色体的遗传信息。该算法促进可行解集中的高质量染色体信息交换,提高了解空间的搜索能力。实验结果说明：改进GA-SVM不仅可决策出与疾病相关的重要特征变量、优化SVM参数,而且可提升分类性能。与前馈BP神经网络及自适应模糊推理系统两种学习算法的比较表明,改进GA-SVM具有更好地表现。

An improved genetic algorithm-support vector machines scheme for coronary artery disease diagnosis

An improved genetic algorithm can automatically determine and simultaneously optimize the feature subsets,parameters of support vector machines（ SVM） in coronary artery disease（ CAD） diagnosis. A novel partitioned multi-gene crossover technique preserves the information in the each gene component and makes offspring to obtain gene information from more than a pair of parent chromosome. The proposed algorithm promotes the information exchange among multiple chromosomes of top ranked solutions,and improves the search ability of the potential region. Experiments results showed that the improved GA can not only determine the critical features related to the CAD,but also improve the classification rate. Comparisons demonstrate that the improved GA-SVM scheme exhibit better performance than the approaches based on the feed forward back-propagation artificial neural network and adaptive neurofuzzy inference systems.