基于多策略融合斑马优化算法的特征选择方法

Method of feature selection based on multiple⁃strategies fusion ZOA

针对传统斑马优化算法在求解复杂优化问题时精度低、收敛速度慢和易陷入局部最优的不足,提出一种多策略融合的改进斑马优化算法(IZOA)。首先,为解决斑马个体初始位置分布不均匀的问题,引入混沌映射来增加探索过程的种群多样性;其次,受自适应权重和黄金正弦算法思想启发,提出一种基于自适应递减权重和黄金正弦更新机制的位置更新策略,用于改进斑马算法的局部寻优与全局探索能力;然后,进行标准测试函数实验,验证了IZOA能够有效提升寻优精度和收敛速度;最后,将K近邻分类器作为待优化目标,选取UCI库的12个标准数据集进行特征选择实验,并利用改进后的算法在特征选择模型中进行最优特征子集搜寻。实验结果表明,相比传统算法,所提算法的平均分类准确率提升4.47%,平均适应度值降低2.5%,验证了该算法在特征选择领域的优越性。

In allusion to the shortcomings of traditional zebra optimization algorithm(ZOA)in solving complex optimization problems,such as low accuracy,slow convergence speed,and susceptibility to local optima,an improved zebra optimization algorithm(IZOA)based on multi-strategies fusion is proposed.In order to improve the uneven initial position distribution of zebra individuals,chaos mapping is introduced to increase the population diversity during the exploration process.Inspired by the ideas of adaptive weighting and the golden sine algorithm(Gold-SA),a position update strategy based on adaptive decending weight and goldensine update mechanism is proposed to improve the local optimization and global exploration capabilities of the zebra algorithm.The experiment of standard test function is conducted to verify that IZOA can effectively improve the optimization accuracy and convergence speed.With K-nearest neighbors(KNN)as the target to be optimized,12 standard data sets of UCI database are selected for feature selection experiments,and the improved algorithm is used to search the optimal feature subset in the feature selection model.The experimental results demonstrate that,in comparison with the original algorithm,the average classification accuracy is improved by 4.47%,and the average fitness value is reduced by 2.5%,validating the superiority of the proposed algorithm in the field of feature selection.