Tiltingtheplayingfield:DynamicalLossfunctionsformachinelearningMiguelRuiz-Garcia12GeZhang1SamuelS.Schoenholz3AndreaJ.Liu1Abstracttionscale(Glorot&Bengio,2010;Xiaoetal.,2018)orlearningrateschedule(H...
ThreeOperatorSplittingwithaNonconvexLossFunctionAlpYurtsever1VarunMangalick1SuvritSra1Abstract2007),incertaintransportandassignmentproblems(Koop-mans&Beckmann,1957;Peyre´etal.,2019),amongcount-Wec...
TheEarthMover’sPinballLoss:QuantilesforHistogram-ValuedRegressionFlorianList1AbstractFigure1.ComicofourmethodforDeepLearning-basedhistogramregression:aneuralnetwork(NN)fθwithweightsθistrainedAlt...
PrincipalBitAnalysis:AutoencodingwithSchur-ConcaveLossSourbhBhadane1AaronB.Wagner1JayadevAcharya1Abstractandgareselectedthroughtrainingfromtheclassoffunc-Weconsideralinearautoencoderinwhichthetions...
LowerBoundsonCross-EntropyLossinthePresenceofTest-timeAdversariesArjunNitinBhagoji1DanielCullina2VikashSehwag3PrateekMittal3Abstractonestablishingfundamentalboundsonlearninginthepres-enceoftest-tim...
LossSurfaceSimplexesforModeConnectingVolumesandFastEnsemblingGregoryW.Benton1WesleyJ.Maddox1SanaeLotfi1AndrewGordonWilson1Abstract✓1<latexitsha1_base64="b2Ff/oUFJw0eznxXS1RygRK2bZk=">AAAB73icbVBNS...
LeveragedWeightedLossforPartialLabelLearning1HongweiWen2JingyiCui1HanyuanHang2JiabinLiu3YisenWang1ZhouchenLin14Abstract2012),etc,andsubsequentlyattractsalotofattentiononmethodologystudies(Fengetal....
MixedCrossEntropyLossforNeuralMachineTranslationHaoranLi1WeiLu1AbstractSourceDieGeburtenrategehtweiterzurück.one-hotmodelgroundpredictiontruthInneuralmachinetranslation,crossentropy(CE)isthestanda...
HowDoesLossFunctionAffectGeneralizationPerformanceofDeepLearning?ApplicationtoHumanAgeEstimationAliAkbari1MuhammadAwais1ManijehBashar2JosefKittler1Abstractconditions,cameraquality,headpose,makeupap...
GeometryoftheLossLandscapeinOverparameterizedNeuralNetworks:SymmetriesandInvariancesBerfinS¸ims¸ek12Franc¸oisGed1ArthurJacot1FrancescoSpadaro1Cle´mentHongler1WulframGerstner2JohanniBrea2Abstrac...
DiscriminativeComplementary-LabelLearningwithWeightedLossYiGao12Min-LingZhang23Abstracttion,includingbutnotlimitedto,semi-supervisedlearning(Chapelleetal.,2006;Oliveretal.,2018;Calderetal.,Compleme...
ConditionalTemporalNeuralProcesseswithCovarianceLossBoseonYoo1JiwooLee2JanghoonJu2SeijunChung1SoyeonKim1JaesikChoi13Abstractducedin(Garneloetal.,2018a),whicharethenoveleffortsthatimplementsuchGPswi...
AsymmetricLossFunctionsforLearningwithNoisyLabelsXiongZhou12XianmingLiu12JunjunJiang12XinGao32XiangyangJi4AbstractToalleviatetheimpactoflabelnoisetoclassifierlearning,onepopularresearchlineistodesi...
SupervisedLearning:NoLossNoCryRichardNock12AdityaKrishnaMenon3Abstractthegoalwastolearninpolynomialtimeafunctionbeingabletopredictalabel(orclass,amongtwopossible)forSupervisedlearningrequiresthespe...
PeerLossFunctions:LearningfromNoisyLabelswithoutKnowingNoiseRatesYangLiu1HongyiGuo2Abstracttensivepriorworksintheliteraturethataimtodevelopal-gorithmstolearnmodelsthatarerobusttolabelnoise(By-Learn...
NormalizedLossFunctionsforDeepLearningwithNoisyLabelsXingjunMa1HanxunHuang1YisenWang2SimoneRomanoSarahErfani1JamesBailey1Abstract2018;Hanetal.,2018b);and4)robustLossfunctionsthatareinherentlytolera...
NeuralArchitectureSearchinAProxyValidationLossLandscapeYanxiLi1MinjingDong1YunheWang2ChangXu1Abstractwhichshowapromisingfutureofthisfield.Althoughhard-wareperformancehasimproved,NASisstillcomputati...
LossFunctionSearchforFaceRecognitionXiaoboWang1ShuoWang1ChengChi2ShifengZhang2TaoMei1AbstractGenerally,theCNNsareequippedwithclassificationLossfunctions(Liuetal.,2017;Wangetal.,2018f;Chenetal.,Infa...
FeatureNoiseInducesLossDiscrepancyAcrossGroupsFereshteKhani1PercyLiang1Abstractdiscrepancy(Liptonetal.,2018)andadverselong-termimpacts(Liuetal.,2018).Therefore,itisimportanttoTheperformanceofstanda...
EliminatingtheInvarianceontheLossLandscapeofLinearAutoencodersRezaOftadeh1JiayiShen1ZhangyangWang1DylanShell1Abstractgenerallythecategorizationofextremepoints,hasflour-ished.Muchworkextendstheresul...