基于粒子群优化的数据分类算法  

Particle Swarm Optimization for Data Classification

作　　者：王旸[1] 刘晓东[1] 徐小慧[2] 胡军 

机构地区：[1]空军工程大学工程学院六系,西安710038 [2]中国人民解放军驻114厂军代表室,西安710077 [3]空军上海军械修理厂,上海200434

出　　处：《系统仿真学报》2008年第22期6158-6162,6168,共6页Journal of System Simulation

摘　　要：设计了一种基于粒子群优化的数据分类算法。新算法首先对数据样本预处理,利用粒子群优化算法通过训练数据进行分类规则的提取,根据提取得到的规则对数据进行分类识别。基于Bayes定理和随机状态转移过程对新算法的收敛性进行分析。通过对UCI数据集分类实验及遥感图像目标识别实验,验证了新算法是一种有效的分类方法。A novel data classification method, Particle Swarm Optimization for classification （PSOC）, was put forward. After pretreating the data, the classification rules were discovered by particle swarm optimization algorithm based on the training samples, and then the data was classified by the discovered rules. The convergence of the new algorithm was analyzed based on Bayes’s theorem and stochastic transform process. Experimental study on UCI machine learning repository and the remote sensing image data shows the proposed algorithm obtains good performances.

关 键 词：分类 粒子群优化 数据挖掘 进化算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]