一种有抗体免疫的云模型数据流聚类算法仿真  

An Antibody Immune Cloud Model Data Stream Clustering Algorithm Simulation

作　　者：邓华军[1] 周士芸[1] 

机构地区：[1]安顺学院物理与电子科学系,贵州安顺561000

出　　处：《科技通报》2013年第9期206-208,共3页Bulletin of Science and Technology

基　　金：贵州省科学技术基金项目(黔科合J字LKA[2012]02号)

摘　　要：云模型是云理论的核心,数据流聚类算法在云模型中有较好的应用前景,但也面临着聚类效率、聚类适应性的难题,为此本文提出了一种有抗体免疫的云模型数据流聚类算法。通过设置加权期望值、熵等参数形成云数据特有的数据概要结构,作为抗体代入人工免疫算法中;利用衰减函数和时刻权重来定量表现不同时刻的数据的重要性程度,并以抗体期望克服率为特征值以维持抗体的多样性,采用淘汰法确保最后的数据概要结构更符合云模型数据流的本质特征。实验表明,该算法在云模型数据流中的聚类处理速度和聚类效率均优于传统算法,具有一定的应用价值。There are good prospects for data stream clustering algorithm used for cloud model as the core of the cloud theory, which also faces the problem of clustering efficiency and clustering suitability, so that a data stream clustering algorithm used in cloud model based on artificial immune principle is proposed in this paper. Cloud data structure is set as antibody-generation of artificial immune algorithm which consists of weighted expectation, entropy and etc. The time instance weight and the attenuation function are used to behave the importance degree of the data of different moments, and antibody expectations overcome rate is made for characteristics value to maintain diversity of antibodies. At last the elimination method is used to ensure that the last synopsis data structure according for the essential characteristics of cloud data stream as more as possible. The experiments show that the algorithm is better than traditional algorithm in both the clustering speed and clustering efficiency and has a certain value.

关 键 词：云模型 数据流聚类 人工免疫原理 数据概要结构 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]