ProvablyStrictGeneralisationBenefitforEquivariantModelsBrynElesedy1SheheryarZaidi2AbstractasSokolicetal.(2017);Sannai&Imaizumi(2019),coveronlytheworstcaseperformanceofalgorithms.TheseworksItiswidel...
LieTransformer:EquivariantSelf-AttentionforLieGroupsMichaelHutchinson1CharlineLeLan1SheheryarZaidi1EmilienDupont1YeeWhyeTeh12HyunjikKim2Abstractducesthenumberofparametersandcomputationalcost.Thisha...
EquivariantNetworksforPixelizedSpheresMehranShakerinava12SiamakRavanbakhsh12AbstractFigure1.Wemodeltherotationalsymmetryofthespherebycom-biningtherotationalsymmetryofaPlatonicsolidandisometriesPixe...
EquivariantLearningofStochasticFields:GaussianProcessesandSteerableConditionalNeuralProcessesPeterHolderrieth1MichaelHutchinson1YeeWhyeTeh12AbstractFigure1.AschematicofEquivariantlearningofvectorfi...
EquivariantMessagePassingforthePredictionofTensorialPropertiesandMolecularSpectraKristofT.Schu¨tt12OliverT.Unke12MichaelGastegger13Abstractoryquicklybecomeprohibitiveforlargesystemsandthepredictio...
E(n)EquivariantGraphNeuralNetworksVictorGarciaSatorras1EmielHoogeboom1MaxWelling1AbstractThispaperintroducesanewmodeltolearngraphneuralnetworksEquivarianttorotations,transla-tions,reflectionsandper...
CyclicallyEquivariantNeuralDecodersforCyclicCodesXiangyuChen1MinYe1Abstractcodes,polarcodes,LDPCcodesandReed-Muller(RM)codes.Neuraldecoderswereintroducedasageneraliza-tionoftheclassicBeliefPropagat...
APracticalMethodforConstructingEquivariantMultilayerPerceptronsforArbitraryMatrixGroupsMarcFinzi1MaxWelling2AndrewGordonWilson1AbstractGroupTypeSoOlvuirnCgoAnlsgtorraiitnhtmEquivariantLinearLayerSy...
UniversalEquivariantMultilayerPerceptronsSiamakRavanbakhsh12Abstractnecessityofparameter-sharingforinvariancewasusedtoprovethelimitationofasinglelayerPerceptron.Thefollow-Groupinvariantandequivaria...
PDO-eConvs:PartialDifferentialOperatorBasedEquivariantConvolutionsZhengyangShen1LingshenHe2ZhouchenLin2JinwenMa1AbstractasignificantadvantageofCNNsisthattheyareshiftequiv-ariant:shiftinganimageandt...
LorentzGroupEquivariantNeuralNetworkforParticlePhysicsAlexanderBogatskiy1BrandonAnderson23JanT.Offermann1MarwahRoussi1DavidW.Miller14RisiKondor256Abstractproblemsmayrequireorbenefitfromlatentspacer...
EquivariantFlows:ExactLikelihoodGenerativeLearningforSymmetricDensitiesJonasKo¨hler1LeonKlein1FrankNoe´123Abstractproblems,suchasvalidconformationsofproteins.Normalizingflowsareexact-likelihoodge...
EquivariantNeuralRenderingEmilienDupont1MiguelAngelBautista2AlexColburn2AdityaSankar2JoshuaSusskind2QiShan2AbstractFigure1.Fromasingleimage(left),ourmodelinfersascenerepresentationandgeneratesnewvi...
AttentiveGroupEquivariantConvolutionalNetworksDavidW.Romero1ErikJ.Bekkers2JakubM.Tomczak1MarkHoogendoorn1AbstractFigure1.Meaningfulrelationshipsamongobjectsymmetries.Thougheveryfigureiscomposedbyth...
GaugeEquivariantConvolutionalNetworksandtheIcosahedralCNNTacoS.Cohen1MauriceWeiler2BerkayKicanaoglu2MaxWelling1AbstractFigure1.AgaugeisasmoothlyvaryingchoiceoftangentframeonasubsetUofamanifoldM.Aga...
EquivariantTransformerNetworksKaiShengTai1PeterBailis1GregoryValiant1Abstractscalingtoeachtrainingimage).WhiledataaugmentationtypicallyhelpsreducethetesterrorofCNN-basedmodels,Howcanpriorknowledgeo...
Cross-Domain3DEquivariantImageEmbeddingsCarlosEsteves1AvneeshSud2ZhengyiLuo1KostasDaniilidis1AmeeshMakadia2AbstractAdesirablepropertyforanimageembeddingisrobustnessto3Dgeometrictransformationsofthe...