超球体检测器覆盖问题的研究  

Research on Coverage of Hypersphere Detectors

作　　者：李鑫鑫[1] 张凤斌[1] 王涛[1] 

机构地区：[1]哈尔滨理工大学计算机学院,黑龙江哈尔滨150080

出　　处：《计算机技术与发展》2008年第6期131-133,137,共4页Computer Technology and Development

基　　金：哈尔滨市学科基金(2003AFXXJ013)

摘　　要：研究表明实值否定选择算法在多维形状空间下呈现出很高的时间和空间复杂性。针对实值否定选择算法中最常采用的超球体检测器,在理论上研究了它的体积,以及体积随半径和维数变化的性质,以此分析了高复杂性出现的原因。针对检测器存在重叠的问题,基于蒙特卡罗方法提出了一个估计检测器覆盖率的算法,用于比较不同检测器生成算法。由于该算法基于随机分布和概率方法,它极大地简化了计算复杂性。The high time and space complexity of real - valued negative selection algorithm in high dimensions has been shown in recent research. Theoretically makes a study of the volume and several properties of hyperspheres detector in real - valued shape - space to analyse the reason of this high complexity. To solve the problem of detectors overlapping, based on Mento Carlo method, proposes an algorithm which estimates the total space （volume） covered by the hyperspheres to compare the coverage of different negative selection algorithms. Using the method of chance distribution and probability, the algorithm reduces the computational complexity.

关 键 词：实值否定选择算法 超球体 检测器覆盖 蒙特卡罗方法 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]