支持向量机研究其在人脸检测中的应用.pdf

Abstract vectormachineisanew method Support on machine—learning statisticS developed shows in the learningtheory．It great suchas advantagesresolving problems and dimensionin small。samples，nonlinear hyper-high domain．The patternrecognition real dataareoften inthe overcome linearlyinseparableinput are space．To this，data into a dimensional mapped feature whichthedata high space，in are and sparse possibly more is separable．In alsonot practice，the kernel mapping explicitly given．Instead，a functionis to the of incorporated theinner simplifycomputation valueofthe product transformed datainthefeature toselectthe space．How forthe appropriateparameters kernelfunction isourresearchwork． quickly Face has recognition