MC-LSTM:Mass-ConservingLSTMPieter-JanHoedt1FrederikKratzert1DanielKlotz1ChristinaHalmich1MarkusHolzleitner1GreyNearing2SeppHochreiter13GünterKlambauer1Abstract(Cohen&Shashua,2017;Gaier&Ha,2019).Th...
ExploringInterpretableLSTMNeuralNetworksoverMulti-VariableDataTianGuo1TaoLin2NinoAntulov-Fantulin1AbstractInadditiontoforecasting,interpretableRNNsaredesir-ableforgaininginsightsintotheimportantpar...
TowardsBinary-ValuedGatesforRobustLSTMTrainingZhuohanLi1DiHe1FeiTian2WeiChen2TaoQin2LiweiWang13Tie-YanLiu2AbstractingproblemofconventionalRNNs,longshort-termmem-ory(LSTM)(Gersetal.,1999;Hochreiter&...
LatentLSTMAllocationJointClusteringandNon-LinearDynamicModelingofSequentialDataManzilZaheer1AmrAhmed2AlexanderJSmola1AbstractFigure1.LLAhasbetterperplexity(lowerisbetter)thanLDAbutmuchfewerparamete...