全部 标题 作者
关键词 摘要

OALib Journal期刊
ISSN: 2333-9721
费用:99美元

查看量下载量

相关文章

更多...

基于生成树的人脸识别

Keywords: 人脸识别,最小生成树,支撑向量机,直方图核

Full-Text   Cite this paper   Add to My Lib

Abstract:

为了提高人脸识别的准确率,研究了基于生成树核的人脸识别.定义了最大生成树的概念,7个生成树核和3个基于直方图核和生成树核的复杂核,并给出了人脸图像的图示,阐述了人脸图像转化成图的过程.实验结果表明,生成树核的性能优于通路核.

References

[1]  SCHOLKOPF B,SMOLA A J.Learning with Kernels[M].Cambridge,Massachusetts London:The MIT Press,2002:3-18.
[2]  VAPNIK V.Statistical learning theory[M].New York:Wiley,1998:1-50.
[3]  HAUSSLER D.Convolutional kernels on discretestructures,UCSC-CRL-99-10[R].Computer ScienceDepartment,UC Santa Cruz,German:Springer Verlag,1999:3-26.
[4]  AIROLA A,PYYSALO S,BJORNE J,et al.All-pathsgraph kernel for protein-protein interaction extraction withevaluation of cross-corpus learning[J].BMC Bioinformatics,BioMed Central,2008:1480-1482.
[5]  AIROLA A,PYYSALO S,BJORNE J,et al.A graphkernel for protein-protein interaction extraction[C]∥Proceedings of the Workshop on Current Trends inBiomedical Natural Language Processing.Association forComputation Linguistics Stroudsburg:[s.n.],2008:1-9.
[6]  KASHIMA H,TSUDA K,INOKUCHI A.Marginalizedkernels between labeled graphs[C]∥Proceedings of the20th International Conference on Machine Learning.Washington,D.C.:[s.n],2003:321-328.
[7]  BORGWARDT K M,KRIEGEL Hans-Peter.Shortest-pathkernels on graphs[C]∥Proceedings of the Fifth IEEEInternational Conference on Data Mining.Houston,TX,USA:[s.n.],2005:74-81.
[8]  RAMON J,GARTNER T.Expressivity versus efficiency ofgraph kernels[C]∥First International Workshop onMining Graphs,Trees and Sequences.Cavtat,Cavtat:[s.n.],2003:65-74.
[9]  HORVATH T,GARTNER T,WROBEL S.Cyclic patternkernels for predictive graph mining[C]∥Proceedings ofthe International Conference on Knowledge Discovery andData Mining.Seattle:[s.n.],2004:158-167.
[10]  GARTNER T,FLACH P,WROBEL S.On graphkernels:Hardness results and efficient alternatives[C]∥In Scholkopf B and Warmuth M,editors,SixteenthAnnual Conference on Computational Learning Theory andSeventh Kernel Workshop,COLT.German:SpringerVerlag,2003:129-143.

Full-Text

Contact Us

service@oalib.com

QQ:3279437679

WhatsApp +8615387084133