摘要
论文提出了一种基于距离贪心策略的灰狼特征选择算法,针对于现今特征选择算法性能较低的劣势进行了改良,该算法用贪心策略改进了原灰狼算法中的位置更新部分,提升了算法开发最优解的能力,改善了收敛速度,并将其应用于心脏病检测实验当中,用于删除多余的特征,用KNN作为分类检测器。实验表明,该算法可以有效提高分类准确度,降低数据特征冗余。
This paper proposes a gray wolf feature selection algorithm based on distance greedy strategy,which is improved for the disadvantages of the current feature selection algorithm.The algorithm improves the position update part of the original gray wolf algorithm with greedy strategy and improves the position.The ability of the algorithm to develop an optimal solution improves the rate of convergence and applies it to heart disease detection experiments to remove redundant features,using KNN as a classifier detector.Experiments show that the algorithm can effectively improve classification accuracy and reduce data feature redundancy.
作者
童坤
钮焱
李军
TONG Kun;NIU Yan;LI Jun(School of Computing,Hubei University of Technology,Wuhan 430068)
出处
《计算机与数字工程》
2020年第4期759-762,792,共5页
Computer & Digital Engineering
基金
湖北省教育厅科学研究计划项目(编号:D2014403)资助。
关键词
灰狼优化算法
特征选择
入侵检测
grey wolf optimization
feature selection
heart disease
