简介:Supportvectormachines(SVMs)haveshownremarkablesuccessinmanyapplications.However,thenon-smoothfeatureofobjectivefunctionisalimitationinpracticalapplicationofSVMs.Toovercomethisdisadvantage,atwicecontinuouslydifferentiablepiecewise-smoothfunctionisconstructedtosmooththeobjectivefunctionofunconstrainedsupportvectormachine(SVM),anditissuesapiecewise-smoothsupportvectormachine(PWESSVM).Comparingtotheothersmoothapproximationfunctions,thesmoothprecisionhasanobviousimprovement.ThetheoreticalanalysisshowsPWESSVMisgloballyconvergent.Numericalresultsandcomparisonsdemonstratetheclassificationperformanceofouralgorithmisbetterthanothercompetitivebaselines.
简介:Supportvectormachines(SVMs)havebeenintensivelyappliedinthedomainsofspeechrecognition,textcategorization,andfaultsdetection.However,thepracticalapplicationofSVMsislimitedbythenon-smoothfeatureofobjectivefunction.Toovercomethisproblem,anovelsmoothfunctionbasedonthegeometryofcircletangentisconstructed.Itsmoothesthenon-differentiabletermofunconstrainedSVM,andalsoproposesacircletangentsmoothSVM(CTSSVM).Comparedwithothersmoothapproachingfunctions,itssmoothprecisionhadanobviousimprovement.TheoreticalanalysisprovedtheglobalconvergenceofCTSSVM.NumericalexperimentsandcomparisonsshowedCTSSVMhadbetterclassificationandlearningefficiencythancompetitivebaselines.
简介:Thesum-ratecapacityofCDMAandTDMAinvectorbroadcastchannelisresearched.Anupperboundofsum-ratecapacityinCDMAsystemandalowerboundofsum-ratecapacityinTDMAsystemarederivedrespectively.Accordingtothesetwobounds,theratioofthetwoboundsatmostcanachievemin(M,K).HereMistheantennanumberofthetransmitterandKisthenumberoftheuserinthisvectorbroadcastchannel.
简介:Asthesignificantbranchofintelligentvehiclenetworkingtechnology,theintelligentfatiguedrivingdetectiontechnologyhasbeenintroducedintothepaperinordertorecognizethefatiguestateofthevehicledriverandavoidthetrafficaccident.Thedisadvantagesofthetraditionalfatiguedrivingdetectionmethodhavebeenpointedoutwhenwestudyonthetraditionaleyetrackingtechnologyandtraditionalartificialneuralnetworks.Onthebasisoftheimagetopologicalanalysistechnology,Haarlikefeaturesandextremelearningmachinealgorithm,anewdetectionmethodoftheintelligentfatiguedrivinghasbeenproposedinthepaper.Besides,thedetailedalgorithmandrealizationschemeoftheintelligentfatiguedrivingdetectionhavebeenputforwardaswell.Finally,bycomparingtheresultsofthesimulationexperiments,thenewmethodhasbeenverifiedtohaveabetterrobustness,efficiencyandaccuracyinmonitoringandtrackingthedrivers’fatiguedrivingbyusingthehumaneyetrackingtechnology.
简介:Withthewideapplicationofvirtualizationtechnologyinclouddatacenters,howtoeffectivelyplacevirtualmachine(VM)isbecomingamajorissueforcloudproviders.Theexistingvirtualmachineplacement(VMP)solutionsaremainlytooptimizeserverresources.However,theypaylittleconsiderationonnetworkresourcesoptimization,andtheydonotconcerntheimpactofthenetworktopologyandthecurrentnetworktraffic.Amulti-resourceconstraintsVMPschemeisproposed.Firstly,theauthorsattempttoreducethetotalcommunicationtrafficinthedatacenternetwork,whichisabstractedasaquadraticassignmentproblem;andthenaimatoptimizingnetworkmaximumlinkutilization(MLU).Ontheconditionofslightvariationofthetotaltraffic,minimizingMLUcanbalancenetworktrafficdistributionandreducenetworkcongestionhotspots,aclassiccombinatorialoptimizationproblemaswellasNP-hardproblem.Antcolonyoptimizationand2-optlocalsearcharecombinedtosolvetheproblem.SimulationshowsthatMLUisdecreasedby20%,andthenumberofhotlinksisdecreasedby37%.
简介:Asthesystemperformanceisobviouslyimprovedbyintroducingtheconceptofrelayintothetraditionalorthogonalfrequencydivisionmultipleaccess(OFDMA)systems,resourceschedulinginrelay-enhancedOFDMAsystemsisworthyofbeingstudiedcarefully.Tosolvetheoptimizationproblemofachievingthemaximumthroughputwhilesatisfyingthequalityofservice(QoS)andguaranteeingthefairnessofusers,anovelresourceschedulingschemewithQoSsupportforthedownlinkoftwo-hoprelay-enhancedOFDMAsystemsisproposed.Theproposedscheme,whichisconsideredbothinthefirsttimesub-slotbetweendirectlinkusersandrelaystations,andthesecondtimesub-slotamongrelaylinkusers,takesQoSsupportintoconsideration,aswellasthesystemthroughputandthefairnessforusers.Simulationresultsshowthattheproposedschemehasgoodperformanceinmaximizingsystemthroughputandguaranteeingtheperformanceintheservicedelayandthedatalossrate.