基于稀疏表示的多重集典型相关分析算法研究-模式识别与智能系统专业论文.docxVIP

研究生优秀毕业论文 硕士学位论文Abstract 硕士学位论文 Abstract Canonical correlation analysis(CCA)is an effective algorithm for feature extraction, the p狁m Which is used in correlation problem between two groups of feature data m rec唧ition．In the real world，the sanle pattern always exhibits various of ch盯acteriStlcs‘ multlset Therefore，this article embarks from the fusion of multi group feature·The 1s the callonical correlation analysis(MCCA)algorithm based on sparse rePresentatlon foundation of the research．Then，we combine with the kernel technology,cla8s set infonnation，maximizing margin criterion and other related basic theory knowledge·We this p印er，the up a series of algorithm，feature extraction and feature fusion·In specltlc research and innovation work includes the following parts： f 1)Due to the canonical correlation analysis algorithm based on sparse preserVlng proiection can not meet the needs of the nonlinear problem in the practlcal work，and correlation algorithm inspired bv kemel technology,we research the canonical a11alysls to av01d the based on kemel sparse preserving projection(KSPCCA)·In order optlmal selection of kemel劬ction in the choose of kernel and fuse multi group feature we research a simple multiset canonical correlation analysis algorithm based the舢ltiple on kemel sparse preserving projection according to the multiple kernel le狃ing idea锄d multiset caIlonical correlation analysis algorithm based on sparse preservmg projec‘10n· f21 We s01ve the regularization equation of the sparse representation coenlclents mm L2 no咖．We can get the collaborative representation coefficients·We establish the s锄ple reconstmcting adjacency with the collaborative representation coefficients to extract doesn’‘ subspace feature．Because the idea of the collaborative representation projectlon consider the class infonnation，then，we add class information into the colIaboratlVe representation projection，we propose the theory of the collaboratiVe representatlon discriminative