KernelMethodsforPatternAnalysis

Kernelmethodsprovideapowerfulandunifiedframeworkforpatterndiscovery,motivatingalgorithmsthatcanactongeneraltypesofdata(e.g.strings,vectorsortext)andlookforgeneraltypesofrelations(e.g.rankings,classifications,regressions,clusters).Theapplicationareasrangefromneuralnetworksandpatternrecognitiontomachinelearninganddatamining.Thisbook,developedfromlecturesandtutorials,fulfilstwomajorroles:firstlyitprovidespractitionerswithalargetoolkitofalgorithms,kernelsandsolutionsreadytouseforstandardpatterndiscoveryproblemsinfieldssuchasbioinformatics,textanalysis,imageanalysis.Secondlyitprovidesaneasyintroductionforstudentsandresearcherstothegrowingfieldofkernel-basedpatternanalysis,demonstratingwithexampleshowtohandcraftanalgorithmorakernelforanewspecificapplication,andcoveringallthenecessaryconceptualandmathematicaltoolstodoso.