摘要
针对网络初始权矢量选取的不确定性问题,提出了粒子群优化-自组织映射（PSO-SOM）算法,利用PSO算法优化SOM网络的初始权矢量,进而进行分类.将提出的方法用于基因表达数据的分类判别中,使得SOM网络的误差平方和大大下降,提高了网络的分类精度,表明PSO-SOM算法用于数据的分类判别是切实有效的.
To solve the problem of uncertainty in the selection of SOM （self-organizing map） networks＇ initial weights, a PSO （particle swarm optimization）-SOM algorithm was proposed. First the PSO algorithm was used to optimize the initial weight vectors of SOM networks, and then the data were clustered by SOM networks. Finally the proposed method was applied to clustering analysis of gene expression data, and the results showed that the sum of squared errors was reduced by using the PSO-SOM algorithm, also the precision of clustering is improved. It is concluded that the PSO-SOM algorithm is efficient to clustering analysis of high dimensional data.
出处
《高技术通讯》
CAS
CSCD
北大核心
2006年第10期1014-1018,共5页
Chinese High Technology Letters
基金
国家自然科学基金（20506003）、教育部科学技术研究重点项目（106073）和上海启明星项目（04QMX1433）资助项目.
