Slide 1TANGO Table ANalysis for Generating OntologiesIntegrating and Storing Uncertain DataSlide 4Slide 5Slide 6Slide 7Slide 8fleck veltergonsity (ld/gg)hepth(gd)burlam 1.2 120falder 2.3 230multon 2.5 400repeat:1. understand table2. generate mini-ontology3. match with growing ontology4. Adjust & mergeuntil ontology developedTANGOTable ANalysis for Generating Ontologies velterhepthgonosityfleck1has1:*1has1:* velterhepthgonosityfleck1has1:*1has1:*TANGO in a nutshell: TANGO repeatedly turns raw tables into conceptual mini-ontologies and integrates them into a growing ontology.GrowingOntologyIntegrating and Storing Uncertain DataBasic Skills:- movement- capability- logical- behavioral…………Clinician: LeeRobot: AzimoAssembleAssembleSafety Layer:Default Layer:Task: Play Imitation GameWant to play a game! Let’s imitate her!Documents Patterns ResultsSortedABCLayoutLogical-Name-Title-CityAaron David …Aarons George S …Abbott Charles H ……W. S. NEWBURYW. H. ADAMSJOSEPH BACHMAN…T. M. GatchE. H. StolteW. S. Newbury…Truly Dynamic Behavior•Graphics•Machine Learning•SimulationTruly Dynamic Behavior•Graphics•Machine Learning•SimulationHuman Behavior•Fire Escape Placement•Fugitive Chase Simulation•Adaptive Fire-fightingHuman Behavior•Fire Escape Placement•Fugitive Chase Simulation•Adaptive Fire-fightingAnimal Learning•Parent-Child Learning Transfer•Predator Introduction•Mutual Genetic AdaptationAnimal Learning•Parent-Child Learning Transfer•Predator Introduction•Mutual Genetic AdaptationMovies and Games•1,000+ Asymmetric Actors•Human-like AI Adaptation•Screen-based IntelligenceMovies and Games•1,000+ Asymmetric Actors•Human-like AI Adaptation•Screen-based IntelligenceGovernment ProfessionalsGovernment ProfessionalsBiologistsBiologistsEntertainment IndustryEntertainment IndustryBrian Ricks - BYU CS - October, 2009Feature ExtractionHappy Bright Peaceful Dark Gloomy? ? ? . . .Happy, Bright, Energetic,?,?,?,?, ...Wet,Happy,?,?,?,...Dark, Gloomy,?,?,?,...Incremental Multi-label Classification with Unknown LabelsWe cannot assume implicitnegativity for images withmissing labels.We don't know which labels we might encounter nor how many labels there will be during training.We need to be able to dynamicallyadd new labels into our learning model.0/9 algorithmsHeuristicsDN=1CL=0.02DS=0.01MV=0Type: