基于小生境策略的检测器生成算法  

An Detector Generation Algorithm Based on Niching Strategy

作　　者：时晨[1] 申普兵[1] 沈向余[1] 于鹏飞 

机构地区：[1]西安通信学院,陕西西安710106 [2]沈阳军区司令部作战指挥数据中心,辽宁沈阳110805

出　　处：《计算机技术与发展》2011年第9期81-84,共4页Computer Technology and Development

基　　金：2009年度通信和指挥自动化装备军内科研项目(TZ-XATY-Y-B-2009-001)

摘　　要：引入小生境思想能够较好地维持解的多样性,同时具有较高的全局寻优能力和收敛速度,特别适合于复杂多峰函数的优化问题。为了提高检测器的生成效率,降低算法开销,引入小生境进化原理用于未成熟检测器的生成;通过计算初始种群个体之间的海明相似度,依据个体之间的亲和力和适应度,构建多个小生境并分别进行进化,提出了基于小生境策略的检测器生成算法;对算法进行了验证,讨论了实验中涉及的参数,与穷举检测器生成算法进行了比较;通过对实验结果的分析,在一定条件下,该算法基本达到了预期的目标。Introducing niching into algorithm could maintain the diversity of results,also have better ability of global optimization and convergence rate,and is more adopted to use in optimization problem of multimodal function.In order to improve the efficiency of generating detectors and reduce the cost of algorithm,it uses niching evolutionary theory in the immature detectors generation.According to the hamming similarity between the strings of initial population and the affinity and fitness of strings,niches are built and evolved respectively,an detector generation algorithm based on niching strategy is proposed.The parameters which involved in simulation experiment are discussed;The algorithm compares with exhaustive detector generation algorithm.Through analysis of experimental results,the algorithm basically reached the expected goal under certain conditions.

关 键 词：小生境 检测器 阴性选择 适应度 亲和力 入侵检测 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]