基于RBF核的SVM學習算法優(yōu)化及其在蛋白質二級結構預測中的應用.pdf

1、武漢科技大學碩士學位論文基于RBF核的SVM學習算法優(yōu)化及其在蛋白質二級結構預測中的應用姓名：李琳申請學位級別：碩士專業(yè)：計算機應用技術指導教師：張曉龍20060430第1I頁武漢科技大學碩士學位論文AbstractThereiSnoauniforillmodelforthechoiceofSVM’SkemelfunctionsanditsparametersThisthesisanalysesGridSearchMethod(GSM)

2、andBilinearSearchMethodmSM)Byimprovingthetwomethods，weproposeanewparametersoptimizationmethodofSVMbasedonRBFkemelBilinearGridSearchMethodfBGSM)BGSMcombinestheadvantagesofGSMandBSM，appearingtobeamoreeffectivelearningalgor

3、ithmOntheconditionofkeepingthelearningaccuracynodescending，themethodgreatlyreducesthenumberoftrainingSVMBGSMhasbeencheckedwithUCIdatasetsTheexperimentsshowthatBGSMhasbetterlearningperformanceandlearningaccuracythanrelate

4、dmethods，ApplicationofBGSMtothefieldofproteinsecondarystructurepredictionalsoobtainsbetterlearningaccuracythanrelatedalgorithms，wheremovingwindowandmultiplealignmentmethodsareusedasencodingmethodsByaligningthesimilarityo

5、funknownsequencesandknownsequences，wecanfindwhetherthereishomologouswiththesesequences，whichgivesUSevolutionaryinformationByimprovingGSMandBSM，weproposeandimplementanoptimizationalgorithmofSVMbasedonRBFkernelByimprovingl