基于改进的蜻蜓算法的特征选择

An Improved Dragonfly Algorithm for Feature Selection

蜻蜓算法是一种近年提出的元启发式优化算法,它主要是模拟自然界中蜻蜓的捕食和迁徙行为。原始的蜻蜓算法跟其他许多群体智能优化算法一样,存在着自身的缺陷,容易陷入局部最优,并且收敛速度较慢。为了提高蜻蜓优化算法的性能,在算法种群初始化阶段引入混沌映射策略,提高了初代种群的质量,并且将原蜻蜓算法的线性惯性权重做了非线性改进,提高了算法的收敛速度,最后运用于特征选择来检验其实际效果。实验结果表明,改进后的蜻蜓算法比原算法的效果更好。

Dragonfly algorithm is a meta heuristic optimization algorithm proposed in recent years,which mainly simulates the predation and migration behavior of dragonflies in nature.The original Dragonfly algorithm,like many other swarm intelligence optimization algorithms,has its own defects,easy to fall into local optimum,and slow convergence speed.In order to improve the performance of dragonfly optimization algorithm,this paper introduces chaos mapping strategy in the initialization phase of the algorithm population,improves the quality of the initial population,and makes nonlinear improvement on the linear inertia weight of the original Dragonfly algorithm.It also improves the convergence speed of the algorithm,and finally applies it to feature selection to test its actual effect.The experimental results show that the improved Dragonfly algorithm is better than the original algorithm.