报告摘要：Theconfluenceofsensing,communicationandcomputingtechnologiesisallowingcaptureandaccesstodata,indiverseformsandmodalities,inwaysthatwereunimaginableevenafewyearsago.Suchdataareakeyenablerofhumanbehavioralresearchanditstranslationalapplications.Importantly,thesedataaffordtheanalysisandinterpretationofmultimodalcuesofverbalandnon-verbalhumanbehavior.Thesedatasourcesnotonlycarrycrucialinformationaboutaperson’sintent,identityandtraitbutalsounderlyingattitudesandemotions.Automaticallycapturingthesecues,althoughvastlychallenging,offersthepromiseofnotjustefficientdataprocessingbutintoolsfordiscoverythatenablehithertounimaginedinsights.Recentcomputationalapproachesthathaveleveragedjudicioususeofbothdataandknowledgehaveyieldedsignificantadvancesinthisregards,forexampleinderivingrich,context-awareinformationfrommultimodalsignalsourcesincludinghumanspeech,language,andvideosofbehavior.Theseareevencomplementedandintegratedwithdataabouthumanbrainandbodyphysiology.Thistalkwillfocusonsomeoftheadvancesandchallengesingatheringsuchdataandcreatingalgorithmsformachineprocessingofsuchcues.ItwillhighlightsomeofourongoingeffortsinBehavioralSignalProcessing(BSP)—technologyandalgorithmsforquantitativelyandobjectivelyunderstandingtypical,atypicalanddistressedhumanbehavior—withaspecificfocusoncommunicative,affectiveandsocialbehavior.ThetalkwillillustrateBehavioralInformaticsapplicationsofthesetechniquesthatcontributetoquantifyinghigher-level,oftensubjectivelydescribed,humanbehaviorinadomain-sensitivefashion.ExampleswillbedrawnfromhealthandwellbeingrealmssuchasAutism,Coupletherapy,DepressionandAddictioncounseling.