简介:Thispaperconcernstheproblemofobjectsegmentationinreal-timeforpickingsystem.Aregionproposalmethodinspiredbyhumanglancebasedontheconvolutionalneuralnetworkisproposedtoselectpromisingregions,allowingmoreprocessingisreservedonlyfortheseregions.Thespeedofobjectsegmentationissignificantlyimprovedbytheregionproposalmethod.Bythecombinationoftheregionproposalmethodbasedontheconvolutionalneuralnetworkandsuperpixelmethod,thecategoryandlocationinformationcanbeusedtosegmentobjectsandimageredundancyissignificantlyreduced.Theprocessingtimeisreducedconsiderablybythistoachievetherealtime.Experimentsshowthattheproposedmethodcansegmenttheinterestedtargetobjectinrealtimeonanordinarylaptop.
简介:Synchronouschipsealisanadvancedroadconstructingtechnology,andthegravelcoveragerateisanimportantindicatoroftheconstructionquality.Inthispaper,anovelapproachforgravelcoverageratemeasurementisproposedbasedondeeplearning.Convolutionalneuralnetwork(CNN)isusedtosegmenttheimageofgroundcoveredwithgravels,andthegravelcoveragerateiscomputedbythepercentageofgravelpixelsinthesegmentedimage.Thegravelcoverageratedatasetformodeltrainingandtestingisbuilt.Theperformanceoffullyconvolutionalneuralnetwork(FCN)andU-Netmodelinthedatasetistested.AbettermodelnamedGravelNetisconstructedbasedonU-Net.Thescaledexponentiallinearunit(SELU)isemployedintheGravelNettoreplacethepopularcombinationofrectifiedlinearunit(ReLU)andbatchnormalization(BN).Dataaugmentationandalphadropoutareperformedtoreduceoverfitting.Theexperimentalresultsdemonstratetheeffectivenessandaccuracyofourproposedmethod.OurtrainedGravelNetachievesthemeangravelcoveragerateerrorof0.35%ontestdataset.
简介:ThemaingoalofroutingsolutionsistosatisfytherequirementsoftheQualityofService(QoS)foreveryadmittedconnectionaswellastoachieveaglobalefficiencyinresourceutilization.InthispaperproposesasolutionbasedonHopfieldneuralnetwork(HNN)todealwithoneofrepresentativeroutingproblemsinuni-castrouting,i.e.themulti-constrained(MC)routingproblem.ComputersimulationshowsthatwecanobtaintheoptimalpathveryrapidlywithournewLyapunovenergyfunctions.
简介:Basedoncurrentresearchonapplicationsofchaoticneuronnetworkforinformationprocessing,thestabilityandconvergenceofchaoticneuronnetworkareprovedfromtheviewpointofenergyfunction.Moreover,anewauto-associativematrixisdevisedforartificialneuralnetworkcomposedofchaoticneurons,thus,animprovedchaoticneuronnetworkforassociativememoryisbuiltup.Finally,theassociativerecallingprocessofthenetworkisanalyzedindetailandexplanationsofimprovementaregiven.
简介:Gatematrixlayoutproblemplaysanimportantroleinintegratedcircuitdesign,butitsoptimizationisNP-hard.Inthispaper,typicalgatelayoutproblemisanalysedandadaptedtoneuralnetworkrepresentation,furthermorethesimulatedresultsaregiven.
简介:Inthisstudy,aMulti-LayerBPneuralnetwork(MLBP)withdynamicthresholdsisemployedtobuildaclassifiermodel.Astothedesignoftheneuralnetworkstructure,theoreticalguidanceandplentifulexperimentsarecombinedtooptimizethehiddenlayers'parameterswhichincludethenumberofhiddenlayersandtheirnodenumbers.Theclassifierwithdynamicthresholdsisusedtostandardizetheoutputforthefirsttime,anditimprovestherobustnessofthemodeltoahighlevel.Finally,theclassifierisappliedtoforecastboxofficerevenueofamoviebeforeitstheatricalrelease.ThecomparisonresultswiththeMLPmethodshowthattheMLBPclassifiermodelachievesmoresatisfactoryresults,anditismorereliableandeffectivetosolvetheproblem.
简介:Thispaperdescribesamodifiedspeed-sensorlesscontrolforinductionmotor(IM)basedonspacevectorpulsewidthmodulationandneuralnetwork.AnElmanANNmethodtoidentifytheIMspeedisproposed,withIMparametersemployedasassociatedelements.TheBPalgorithmisusedtoprovideanadaptiveestimationofthemotorspeed.Theeffectivenessoftheproposedmethodisverifiedbysimulationresults.TheimplementationonTMS320F240fixedDSPisprovided.
简介:Thispaperpresentsanewsolutiontotheimagesegmentationproblem,whichisbasedonfuzzy-neural-networkhybridsyste
简介:Choosingtherightcharacteristicparameteristhekeytofaultdiagnosisinanalogcircuit.Thefeatureevaluationandextractionmethodsbasedonneuralnetworkarepresented.Parameterevaluationofcircuitfeaturesisrealizedbytrainingresultsfromneuralnetwork;thesuperiornonlinearmappingcapabilityiscompetentforextractingfaultfeatureswhicharenormalizedandcompressedsubsequently.Thecomplexclassificationproblemonfaultpatternrecognitioninanalogcircuitistransferredintofeatureprocessingstagebyfeatureextractionbasedonneuralnetworkeffectively,whichimprovesthediagnosisefficiency.Afaultdiagnosisillustrationvalidatedthismethod.
简介:Handwrittensignaturerecognitionispresentedbasedonananglefeaturevectorbyusingtheartificialneuralnetwork(ANN)inthisresearch.Eachsignatureimagewillberepresentedbyananglevector.ThefeaturevectorwillconstitutetheinputtotheANN.Thecollectionofsignatureimagesisdividedintotwosets.OnesetwillbeusedfortrainingtheANNinasupervisedfashion.TheothersetwhichisneverseenbytheANNwillbeusedfortesting.Aftertraining,theANNwillbetestedbyrecognizingthesignatures.Whenasignatureisclassifiedcorrectly,itisconsideredcorrectrecognition,otherwiseitisafailure.Theachievedrecognitionrateofthissystemis94%.
简介:Performancerobustnessproblemsviathestatefeedbackcontrollerareinvestigatedforaclassofuncertainnonlinearsystemswithtime-delayinbothstateandcontrol,inwhichtheneuralnetworksareusedtomodelthenonlinearities.Byusinganappropriateuncertaintydescriptionandthelineardifferenceinclusiontechnique,sufficientconditionsforexistenceofsuchcontrollerarederivedbasedonthelinearmatrixinequalities(LMIs).UsingsolutionsofLMIs,astatefeedbackcontrollawisproposedtostabilizetheperturbedsystemandguaranteeanupperboundofsystemperformance,whichisapplicabletoarbitrarytime-delays.
简介:Duetothedemandofdataprocessingforpolariceradarinourlaboratory,aCurveletThresholdingNeuralNetwork(TNN)noisereductionmethodisproposed,andanewthresholdfunctionwithinfinite-ordercontinuousderivativeisconstructed.ThemethodisbasedonTNNmodel.InthelearningprocessofTNN,thegradientdescentmethodisadoptedtosolvetheadaptiveoptimalthresholdsofdifferentscalesanddirectionsinCurveletdomain,andtoachieveanoptimalmeansquareerrorperformance.Inthispaper,thespecificimplementationstepsarepresented,andthesuperiorityofthismethodisverifiedbysimulation.Finally,theproposedmethodisusedtoprocesstheiceradardataobtainedduringthe28thChineseNationalAntarcticResearchExpeditionintheregionofZhongshanStation,Antarctica.Experimentalresultsshowthattheproposedmethodcanreducethenoiseeffectively,whilepreservingtheedgeoftheicelayers.