摘要
对于海量和高维的大规模数据聚类问题,其数据个数以及模式种类通常处于一个动态增加的过程之中,为此进行增量、并行算法的设计,以提供更好的计算能力是十分必要的.注意到人脑增量学习的本质和RSOM(Re-cursive Self-Organizing Map)的层次化、分布式结构特点,本文研究了基于高性能集群并行计算环境的增量、分布式RSOM并行算法,并以视频图像特征集实例证实了算法的可行性.
For large data-set with high dimeusionality, of which the numbers of samples and patterns increase dynamically, in roder to improve the computing-efficiency, it is necessary to design parallel incremental clustering algorithm. Noticing the nature of the human brain-an incremental studying style, and the hierarchical and distributed structure properties of a RSOM tree, a Cluster- computer system based incremental and distributed parallel algorithm of RSOM tree is proposed. The performance of this method is tested with the large feature data sets which are extracted from a large amount of video pictures.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2007年第3期385-391,共7页
Acta Electronica Sinica
基金
国家863计划(No.2003AA134030)
国家重点实验室基金项目(No.9140C8001020603)