【摘要】 为解决数据维数高、信息冗余导致的数据处理问题,提出基于改进混沌粒子群的最优特征提取方法。引入初始潜能的概念优化粒子群的初始化,降低传统的随机初始化导致的盲目性;在此基础上同时考虑粒子的适应度值和位置两个影响因素,对权重做出动态调整,调节空间范围搜索的能力,通过早熟判断机制,及时引入混沌变量避免局部最优。KDDCUP99数据实验得到的分类正确率验证了该方法的有效性和高效性。更多还原

【Abstract】 To solve data processing problem caused by high dimensionality of data and information redundancy,optimal features extraction method was proposed based on improved chaos particle swarm.Firstly,the particle position was initialized through introducing the concept of initial potential,and the blindness caused by traditional random initialization was reduced.On this basis,the influence of two factors of fitness values and location was taken into account,and weight was adjusted dynamically to adjust the scope of ability to search space.Then chaotic variables were introduced by the premature judgment mechanism to avoid local optima.The classification accuracy of KDDCUP99 data obtained experimentally verifies the effectiveness and efficiency of the presented method.更多还原