摘要
Anintelligentmethodforimprovingpositionlinearityofposition-sensitivedetector(PSD),basedonsupportvectormachines(SVMs),isdeveloped.TheSVMisestablishedbasedonthestructuralriskminimizationprincipleratherthanminimizingtheempiricalerrorcommonlyimplementedinneuralnetworks.SVMcanachievehighergeneralizationperformance.TrainingSVMisequivalenttosolvingalinearlyconstrainedquadraticprogrammingproblem,thusthesolutionofSVMisalwaysuniqueandgloballyoptimal.Theimprovingpositionlinearityprocedurehasbeenillustratedusingatwo-dimensional(2D)PSD.ItispointedoutthatthepositionlinearityofthemeasuringsystemwithaproperSVMcorrectionisimprovedbytwoordersofmagnitudeinthemeasurementrange.
出版日期
2005年04月14日(中国期刊网平台首次上网日期,不代表论文的发表时间)