摘要
针对常用分类方法分类精度较低和内存消耗较高的问题,设计一种基于多吸引子元胞自动机(MACA)的模式分类器tsPCM,把它应用于分布式数据挖掘。通过改变MACA的描述方法,用依赖串和依赖向量将分类过程设计成两阶段,用遗传算法优化设计。实验结果表明tsPCM具有较高的分类精度和较低的内存消耗,分类复杂度由O(n3)降低到线性级O(n),具有较好的应用价值。
Owing to low classification accuracy and high memory overhead of the normal classifier, a Pattern Classifying Machine(PCM) named tsPCM based on Multiple Attractor Cellular Automata(MACA) for Distributed Data Mining(DDM) is designed, by changing the characterization of a MACA to two stage with two linear operators of Dependency String(DS) and Dependency Vector(DV), and employing Genetic Algorithm(GA) formulation. Plentiful experimental results prove the potential of tsPCM. Its classification complexity is declined from O(n^3) to O(n), and it has the respect to excellent classification accuracy and low memory overhead established the availability of the classifier to manipulate the distributed data mining.
出处
《计算机工程》
CAS
CSCD
北大核心
2009年第16期180-182,共3页
Computer Engineering
基金
国家"863"计划基金资助项目"分布式企业信息系统中的复杂授权机制研究"(2003AA414031)
江苏省科技攻关计划基金资助项目"面向成套电气大型企业集团的数字化综合集成系统开发与应用"(BE2007071)
关键词
多吸引子元胞自动机
模式分类器
分布式数据挖掘
Multiple Attractor Cellular Automata(MACA)
Pattern Classifying Machine(PCM)
Distributed Data Mining(DDM)