大数据典型相关分析的云模型方法Canonicalcorrelationanalysisof.PDFVIP

第34 卷第10 期 通 信 学 报 Vol.34 No. 10 2013 年10 月 Journa l on Communications October 2013 doi:10.3969/j .issn.1000-436x.2013.10.015 大数据典型相关分析的云模型方法 杨静，李文平，张健沛 （哈尔滨工程大学 计算机科学与技术学院，黑龙江 哈尔滨 15000 1） 摘 要：针对传统大数据典型相关分析 （CCA, canonical correlation an alys is ）方法的高复杂度在面临大数据PB 级 数据规模时不再适应的现状，提出了一种基于云模型的大数据 CCA 方法。该方法在云计算架构的基础上，通过 云运算将各端点云合并为中心云，并据此产生中心云滴，以中心云滴作为大数据的不确定性复原小样本，在其上 施以CCA 运算，中心云滴的较小数据量提高了运算效率。在真实数据集上的实验结果验证了该方法的有效性。 关键词：大数据；典型相关分析；云模型；云运算；云计算 中图分类号：TP391 文献标识码：B 文章编号：1000-436X (20 13) 10-012 1-14 Canonical correlation analysis of big data based on cloud model YAN G Jing, LI Wen-ping, ZHAN G Jian-pei (College o f Comp uter Sc ience and Techno lo gy, Harb in Engineer ing Univers ty, Harb in 15000 1, China) Abstract: The comp lexity of traditional CCA methods is too high meet the requ irements to analyze b ig data due to their hu ge scale which is reaching the lev el of p eta-byte. A novel approach to CCA was prop osed to mine the b ig data by introducing the cloud model which is a brand-nowel theory ab out the uncertainty artific ial intelligence. A distributed ar- chitecture b ased on cloud computing was estab lished . All of the clouds distributing on the nodes of the distributed archi- tecture wer e comb ined to a center cloud via cloud op eration (whe cloud is a synop sis of data and which is a concept coming from the cloud theory). A typ e of virtual samp le of data called cloud drops created b ased on the center cloud . Fi- nally the computing of CCA was imp osed on the cloud drop s. Th