摘要
结合精英遗传算法"优胜"和稳态遗传算法"劣汰"的优点,提出一种先全局大范围搜索后局部重点搜索的分级遗传算法并用于心电信号的特征选择。针对传统遗传算法易陷入局部极小的问题,提出新的存优去劣扩空间选择算子,使种群中的优良个体保持到下一代,且能淘汰劣质个体,加入新的个体,保证算法可以在全空间搜索;引入拼接算子和切断算子在局部空间搜索,解决了遗传算法收敛速度慢的问题。以朴素贝叶斯分类器分类性能作为特征子集评价标准,在MIT-BIH数据库上的实验结果表明,算法得到的特征子集具有良好的分类性能。
Based on survival of the fittest which is combined elitist strategy of elitist genetic algorithm （GA） with steady GA, an improved elitist genetic algorithm is proposed for feature selection, which could search in a global wide range at first and focus search on local key areas in the latter stage. In order to avoid falling into local optimum, a new selecting operator named ＇select superior, eliminate inferior and enlarge space＇ is put forword, which could not only maintain excellent individuals to next generation, but also eliminate inferior individuals, join new individuals into population so as to search in the whole space. Splicing and cutting operators is introduced in the local search space accelerated the convergence speed. The performance of Naive Bayesian classifier as a feature subset evaluation criteria, the experimental results of the entire MIT-BIH arrhythmia database demonstrate that the feature subset which are obtained from the proposed algorithm has good performance compared with the other algorithms.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第5期1792-1796,共5页
Computer Engineering and Design
基金
天津市应用基础与前沿技术研究计划重点基金项目(11JCZDJC15700)
河北省自然科学基金项目(F2013202104)
关键词
精英策略
遗传算法
特征选择
心电信号
朴素贝叶斯分类器
elitist strategy
genetic algorithm （GA）
feature selectiom ECG
naive Bayesian classifier （NBC）
