Office of the Administrator Awarded Contracts - Machine Learning | Federal Compass

Office of the Administrator Awarded Contracts - Machine Learning

With the most comprehensive market intelligence platform,
we have Office of the Administrator machine learning contracts covered.

Let's Get Started Today
Cloud Program Management Systems Engineering Machine Learning Cyber Security Veteran Owned Woman Owned Total Small Business
1 - 9 of 9
NNX13AN82H - DEVELOPING MACHINE LEARNING ALGORITHMS TO ACCESS BEDROCK AND INTERNAL LAYERS IN POLAR RADAR IMAGERY THIS PROPOSAL IS FOR DEVELOPING MACHINE LEARNING ALGORITHMS IN ORDER TO CLASSIFY BEDROCK AND INTERNAL LAYERS IN POLAR RADAR IMAGERY. THIS RESEARCH WILL PROVIDE ROBUST SCALABLE ALGORITHMS FOR AUTOMATIC ANALYSIS OF THE RAPIDLY GROWING AMOUNT OF POLAR REGION RADAR DATA. KNOWLEDGE OF THE BEDROCK AND INTERNAL LAYERS IS ESSENTIAL TO UNDERSTANDING THE PHYSICS OF RAPID CHANGES IN THE ICE SHEETS AND TO DEVELOP IMPROVED MODELS TO ACCURATELY PREDICT THE ICE SHEETS' CONTRIBUTION TO SEA LEVEL RISE. METHODS/TECHNIQUES: DEVELOPING MACHINE LEARNING ALGORITHMS FOR THE APPLICATION OF IDENTIFYING LAYER BOUNDARIES WILL REQUIRE USING ICE THICKNESS AND ACCUMULATION IMAGE DATA COLLECTED IN GREENLAND AND ANTARCTICA. FOR THESE IMAGES, LABELED LAYER BOUNDARIES WILL SERVE AS GROUND TRUTH FOR THE TRAINING MODEL. THE TRAINING DATA WILL BE USED TO DISCOVER PATTERNS AND RELATIONSHIPS NEEDED FOR THE INVESTIGATED MACHINE LEARNING ALGORITHMS. ONCE FEATURES HAVE BEEN EXTRACTED. I WILL EXPLORE VARYING LEARNING ALGORITHMS FOR DEVELOPING A MODEL TO CLASSIFY INSTANCES IN THE POLAR RADAR DATA. IF NECESSARY, I WILL ALSO EXPLORE THE USE OF MULTIPLE, HETEROGENEOUS LEARNING ALGORITHMS FOR COLLECTIVELY CLASSIFYING BEDROCK AND INTERNAL LAYERS SINCE SINGLE ALGORITHMS MAY PRODUCE LOW LEARNING RATES BECAUSE OF THE NOISY RADAR DATA. THESE PROCEDURES WILL CONCLUDE WITH POST ANALYSIS IN ORDER TO VERIFY CORRECTNESS FOR RELEASING TO THE SCIENTIFIC COMMUNITY. SIGNIFICANCE: IT IS WELL ACCEPTED IN THE SCIENTIFIC COMMUNITY THAT GLOBAL CLIMATE CHANGE IS ONGOING. ALSO, RECENT OBSERVATIONS SHOW THAT PARTS OF THE POLAR ICE SHEETS IN GREENLAND AND WEST ANTARCTICA ARE THINNING RAPIDLY. TO FULLY UNDERSTAND THE IMPACT OF GLOBAL CLIMATE CHANGE ON THE ICE SHEETS, IT IS NECESSARY COMPLETELY THE PROCESSES CONTROLLING THE RAPID INTEGRATION AND RETREAT OF KEY GLACIERS. A FEW OF THE MOST IMPORTANT DATASETS ARE ICE THICKNESS AND MAPPED INTERNAL LAYERS IN POLAR FIRM. THESE DATA, HOWEVER, REQUIRE ACCURATE SELECTION OF LAYER BOUNDARIES FOR DEVELOPING NUMERICAL MODELS OF THE ICE SHEETS. UNDERSTANDING THE BEDROCK AND INTERNAL LAYERS IS DIRECTLY RELATED TO THE OBJECTIVES OF THE NASA EARTH AND SPACE SCIENCE FELLOWSHIP IN THE FOCUS OF CLIMATE VARIABILITY AND CHANGE AS THE EARTH SURFACE AND ITS INTERIOR.
Grant for Research
Contractor
TRUSTEES OF INDIANA UNIVERSITY
Contracting Agency/Office
National Aeronautics and Space Administration»Mission Support Directorate»NASA Shared Services Center
Effective date
08/06/2013
Obligated Amount
$89.8k
NNX13AN63H - THIS STUDY WILL INVESTIGATE FUNDAMENTAL QUESTIONS REGARDING THE MODES OF CLIMATE VARIABILITY AND THEIR IMPACT ON REGIONAL PRECIPITATION IN BOTH SPACE AND TIME BY APPLYING OBJECT-ORIENTED CONNECTIVITY ALGORITHM AND ADVANCED MACHINE LEARNING CLUSTERING ALGORITHMS TO NASA'S TRMM PRECIPITATION DATA, NASA'S WATER VAPOR PROJECT NVAP-M TOTAL PRECIPITABLE WATER (TPW) AND NASA'S MERRA RETROSPECTIVE DATA. THIS APPROACH ALLOWS US TO ORGANIZE PRECIPITATION EVENTS (OR THE ASSOCIATED WATER VAPOR FIELDS) CAPTURED BY NASA SATELLITE PLATFORMS INTO SEPARATE OBJECTS (SEE FIGURE 2). THE ABILITY TO SEPARATE EVENTS INTO OBJECTS CREATES THE OPPORTUNITY TO BUILD A SEARCHABLE DATABASE THAT CAPTURES THE DYNAMICAL STRUCTURE OF THE PRECIPITATION EVENT AND AGGREGATES THE EMPIRICAL CHARACTERISTICS OF THE EVENT (E.G. VOLUME, MAXIMUM PRECIPITATION INTENSITY, AVERAGE INTENSITY, DURATION, AVERAGE SPEED, CENTROID (LATITUDE, LONGITUDE)). ONCE THESE OBJECTS ARE ADDED TO A SEARCHABLE DATABASE, A MACHINE LEARNING APPROACH, THE GAUSSIAN MIXTURE MODEL (GMM), WILL BE USED TO ANALYZE THE EVENT'S CHARACTERISTICS, HELPING TO DEVELOP NEW HYPOTHESES ON GOVERNING CLIMATE PROCESSES AND/OR PATTERNS THAT DETERMINE THE OBJECT'S CHARACTERISTICS. USING THE RESULTS OF THE GMM CLUSTERING, AN ANALYSIS WILL THEN BE PERFORMED TO EXPLORE THE DIFFERENT CLUSTERS' PARAMETERS (MEAN AND COVARIANCE MATRIX) TO DETECT CHANGES IN PRECIPITATION CHARACTERISTICS THAT CORRESPOND TO CLIMATE PHENOMENA, SUCH AS THE EL NINO SOUTHERN OSCILLATION (ENSO) WARM AND COLD PHASES, MADDEN JULIAN OSCILLATION (MJO), ARCTIC OSCILLATION (AO) AND THE NORTH ATLANTIC OSCILLATION (NAO). THIS METHODOLOGY WILL BE EXTENDED AND APPLIED TO NASA MERRA PRECIPITATION ESTIMATES. THE NASA MERRA ANALYSIS, WHICH STARTS IN 1979, ALLOWS US TO ACCESS MORE DATA FOR EXTENSIVE STATISTICAL ANALYSIS. THE REGION OF INITIAL FOCUS FOR THIS STUDY IS THE WESTERN UNITED STATES; SPECIFICALLY WE WILL REVIEW THE VARIABILITY OF PRECIPITATION OVER CALIFORNIA. TO VALIDATE THE MODEL RESULTS, CROSS-VALIDATION WILL BE USED TO TEST THE ROBUSTNESS OF THE MIXTURE MODEL PARAMETER ESTIMATION IN ORDER TO DETERMINE AND BUILD CONFIDENCE IN THE PROPOSED MODEL AND METHODOLOGY. THE NUMBER OF MIXTURE MODEL COMPONENTS WILL BE DETERMINED BY USING AKAIKE INFORMATION CRITERION (AIC) AND BAYESIAN INFORMATION CRITERION (BIC).
Training Grant
Contractor
University of California (UNIVERSITY OF CALIFORNIA, IRVINE)
Contracting Agency/Office
National Aeronautics and Space Administration»Mission Support Directorate»NASA Shared Services Center
Effective date
07/31/2013
Obligated Amount
$60k
NNX13AB65G - INCREASES IN HOT EXTREMES (~HEAT WAVES) ARE AMONG THE MOST LIKELY, TANGIBLE, AND ECOLOGICALLY SIGNIFICANT (MEEHL AND TEBALDI 2004, SMITH 2011) OBSERVED AND EXPECTED FEATURES OF THE EARTH'S CHANGING CLIMATE. IN RECENT YEARS, PROMINENT HEAT WAVES HAVE OCCURRED IN EUROPE (ZAITCHIK ET AL. 2006), AUSTRALIA (TONG ET AL. 2010), AND THE UNITED STATES (NOAA 2011). HOWEVER, PREDICTING THE IMPACT OF HOT EXTREMES IS CHALLENGING. THE SPATIAL DISTRIBUTION OF TEMPERATURES IS HETEROGENEOUS, ESPECIALLY IN COMPLEX TERRAIN (FRIDLEY 2009). ORGANISMS HAVE DIFFERENT SENSITIVITIES BASED ON THEIR BEHAVIOR AND PHYSIOLOGY, AND THEIR RESPONSES MAY BE SUBJECT TO THRESHOLD EFFECTS OPERATING AT FINE TIME SCALES. AMONG TAXA, BIRDS ARE HIGHLY SENSITIVE TO HOT EXTREMES, AND PARTICULARLY SO IN ARID AND MONTANE ECOREGIONS WHERE LARGE-SCALE DIE-OFFS AND ABUNDANCE CHANGES HAVE BEEN OBSERVED (ALBRIGHT ET AL. 2010A, MCKECHNIE AND WOLF 2010). THUS, CHARACTERIZATION OF BOTH THE OCCURRENCE AND THE EFFECTS OF HOT EXTREMES IN THESE LOCATIONS AT RELEVANT SPATIO-TEMPORAL SCALES IS OF PRIME IMPORTANCE. AND BECAUSE THE FREQUENCY AND MAGNITUDE OF HOT EXTREMES ARE LIKELY TO INCREASE FASTER THAN MEAN TEMPERATURES (STERL ET AL. 2008), CONSIDERATION OF FUTURE DYNAMICS AND EFFECTS IS IMPORTANT FOR MANAGEMENT AND CONSERVATION OF ECOSYSTEM PROCESSES AND BIODIVERSITY. THIS RESPONDS TO NASA'S OBJECTIVE TO "ADVANCE EARTH SYSTEM SCIENCE TO MEET THE CHALLENGES OF CLIMATE AND ENVIRONMENTAL CHANGE." CARRIED OUT IN THE DIVERSE DESERT AND MOUNTAIN SYSTEMS OF THE US SOUTHWEST, THERE ARE FOUR SCIENTIFIC AND EDUCATIONAL OBJECTIVES: I: TO CHARACTERIZE MAXIMUM SURFACE AND AMBIENT TEMPERATURES FOR ECOLOGICAL APPLICATIONS IN COMPLEX TERRAIN THROUGH THE INTEGRATION OF REMOTE SENSING, GEOSPATIAL DATA, AND IN SITU TEMPERATURE OBSERVATIONS. II: TO QUANTIFY AND UNDERSTAND THE EFFECTS OF HETEROGENEOUS HOT EXTREMES ON AVIAN SURVIVAL, BEHAVIOR, ABUNDANCE AND COMMUNITY DYNAMICS, AND RANGE LIMITS AT MULTIPLE SPATIO-TEMPORAL SCALES. III: TO ASSESS HOW SPATIO-TEMPORAL VARIABILITY IN HOT EXTREMES MAY SHIFT WITH CLIMATE CHANGE AND ALTER AVIAN SURVIVAL, BEHAVIOR, ABUNDANCE AND COMMUNITY DYNAMICS, AND RANGE LIMITS. IV: TO INCREASE INTEREST IN STEM TOPICS AMONG DISADVANTAGED COLLEGE-BOUND YOUTH BY ENGAGING THEM IN LEARNER-LED INQUIRY-BASED ACTIVITIES INTEGRATING MICROMETEOROLOGY, ORNITHOLOGY, AND NASA IMAGERY. IN OBJECTIVE I, WORK WILL BE BASED ON NASA EARTH OBSERVATION DATA (ESPECIALLY DAILY LAND SURFACE TEMPERATURE) COUPLED WITH EXISTING AND NEWLY DEVELOPED NETWORKS OF GROUND MEASUREMENTS. WE WILL EMPLOY PROCESS-BASED, STATISTICAL, AND MACHINE-LEARNING TECHNIQUES TO RECONSTRUCT GRIDDED TIME SERIES, SUMMARY, AND ANOMALY DATASETS OF AMBIENT AND SURFACE TEMPERATURE AT GRAINS RANGING FROM 10 TO 250 M. IN OBJECTIVE II, WE WILL EXAMINE THE IMPLICATIONS OF HOT EXTREMES FOR A PARTICULARLY SENSITIVE TAXONOMIC GROUP: BIRDS. FIRST WE MODEL VARIATIONS IN SURVIVAL TIMES ACROSS SPACE, TIME, AND AVIAN FUNCTIONAL TRAITS USING PHYSIOLOGICAL MODELING AND DATA. THEN WE CONFRONT PREDICTIONS OF AVIAN RESPONSES AT BEHAVIORAL, COMMUNITY, AND DISTRIBUTIONAL SCALES WITH OBSERVATIONS AND CITIZEN SCIENCE DATASETS. IN OBJECTIVE III, WE WILL USE DYNAMICALLY AND STATISTICALLY DOWNSCALED REGIONAL CLIMATE MODELS TO EXPLORE FUTURE SHIFTS IN THE DYNAMICS OF HOT EXTREMES AND USE MODELS TO EXAMINE FUTURE IMPLICATIONS FOR BIRDS. AND IN OBJECTIVE IV, WE WILL CONDUCT A SERIES OF WORKSHOPS AND FOLLOW-UP ACTIVITIES WITH COHORTS OF COLLEGE-BOUND DISADVANTAGED YOUTH. A CENTRAL ACTIVITY INVOLVES STUDENTS DESIGNING MICROMETEOROLOGICAL EXPERIMENTS IN THE FIELD USING LOW-COST TEMPERATURE SENSORS/LOGGERS AND LINKING OBSERVATIONS TO NASA IMAGERY. THIS PROJECT BUILDS FROM PREVIOUS NASA-FUNDED RESEARCH BY THE PI AND ENGAGES ADDITIONAL EXPERTS IN ECOPHYSIOLOGY (WOLF), AVIAN CONSERVATION BIOLOGY (PIDGEON), REMOTE SENSING (NETELER), REGIONAL CLIMATE MODELING (MEJIA), AND STEM EDUCATION (EWING-TAYLOR) AS COLLABORATORS OR AS CO-I.
Grant for Research
Contractor
NEVADA SYSTEM OF HIGHER EDUCATION (UNIVERSITY OF NEVADA, RENO)
Contracting Agency/Office
National Aeronautics and Space Administration»Mission Support Directorate»NASA Shared Services Center
Effective date
11/29/2012
Obligated Amount
$350k
NNX12AN07G - THE OVERALL GOAL OF OUR PROPOSED RESEARCH IS THE CONTINUING PROTOTYPE DEVELOPMENT OF A FRAMEWORK FOR ESTIMATING LOCAL-SCALE CARBON STOCKS AND FUTURE CARBON SEQUESTRATION POTENTIAL FOR THE STATE OF MARYLAND USING REMOTE SENSING AND ECOSYSTEM MODELING. SPECIFICALLY, WE WILL ADDRESS THE FOLLOWING OBJECTIVES: (1) IMPROVE OUR EXISTING METHODOLOGY FOR CARBON STOCK ESTIMATION AND UNCERTAINTY AND ASSESS ITS EFFICACY ACROSS AN EXPANDED RANGE OF ENVIRONMENTAL AND FOREST CONDITIONS; (2) PROVIDE LOCAL-SCALE ESTIMATES OF CARBON STOCKS AND THEIR UNCERTAINTIES FOR THE ENTIRE STATE OF MARYLAND REPRESENTING EASTERN U.S. FOREST TYPES; (3) INITIALIZE AND RUN A PROGNOSTIC ECOSYSTEM MODEL TO ESTIMATE CARBON STOCKS AND THEIR CHANGES, AND TO ESTIMATE CARBON SEQUESTRATION POTENTIAL; (4) PROVIDE DETAILED VALIDATION OF NATIONAL BIOMASS MAPS USING FIA DATA AND LOCAL-SCALE BIOMASS MAPS. (5) DEMONSTRATE NEW DATA ACQUISITION TECHNOLOGY (SINGLE PHOTON COUNTING) FOR LOW-COST, RAPID CARBON ASSESSMENTS. OUR PROPOSED WORK WILL GREATLY EXPAND OUR COVERAGE FROM 2 TO 24 MARYLAND COUNTIES AND EXTENDS FROM THE TIDEWATER FORESTS OF THE CHESAPEAKE BAY THROUGH THE COASTAL PLAINS AND UPLANDS, TO THE MOUNTAINOUS FORESTS OF WESTERN MARYLAND AND THE APPALACHIANS. THIS GRADIENT IN LAND USE, TOPOGRAPHIC, EDAPHIC, AND CLIMATIC CONDITIONS ENABLES AN APPROPRIATE EXPANSION OF METHODS, MODELS, DATA, AND ASSESSMENTS CONSISTENT WITH THE GOALS OF THE SECOND PHASE OF CMS. OUR OBJECTIVES BUILD FROM OUR PHASE 1 WORK AND LEAD TO A CLEAR SET OF TASKS FOR THE PROPOSED EFFORT. THESE ARE DIVIDED INTO SEVEN CATEGORIES OF ACTIVITIES TRACEABLE TO THIS FRAMEWORK: (1) REMOTE SENSING DATA ACQUISITION AND PROCESSING; (2) FIELD DATA COLLECTION AND ANALYSIS; (3) ALGORITHM DEVELOPMENT AND IMAGE PROCESSING; (4) STATISTICAL AND MACHINE LEARNING MODEL DEVELOPMENT; (5) COUNTY BIOMASS AND UNCERTAINTY MAP GENERATION, AND END-TO-END ERROR ANALYSIS; (6) PROGNOSTIC ECOSYSTEM MODELING, AND; (7) NATIONAL BIOMASS MAP VALIDATIONS. AN ADDITIONAL ELEMENT OF OUR PROPOSED WORK IS A COORDINATED OUTREACH EFFORT TO COUNTY AND STATE AGENCIES TO INFORM AND PROMOTE THEIR ACTIVITIES IN CMS AND INCLUDES A TRANSFER OF TECHNOLOGY TO THE STATE OF VERMONT. TO PROMOTE THIS OUTREACH WE WILL ALSO IMPLEMENT A NEW, WEB-BASED DATA VISUALIZATION, QUERY AND DELIVERY SYSTEM, GRID^INTEL ONLINE (GIO) THAT ALLOWS ANY USER TO CALL UP LIDAR DATA, ASSOCIATED IMAGERY, BIOMASS AND ERROR ESTIMATES FOR ARBITRARY MAP AREAS. DELIVERABLES FOR THIS PROJECT EXPAND UPON THOSE FROM PHASE 1. IN ADDITION TO THE DEVELOPED FRAMEWORK THE PROJECT WILL PRODUCE THE FOLLOWING CMS PRODUCTS: (1) TILED AND MOSAICKED CANOPY HEIGHT AND FOREST/NON-FOREST MAPS AT 2 M AND 30 M RESOLUTION FOR MARYLAND; (2) AGBM MAPS AT 30 M RESOLUTION WITH ASSOCIATED UNCERTAINTY MAPS; (3) ED-MODEL BASED CARBON AND CARBON-FLUX MAPS AT 90 M RESOLUTION; (4) ED-MODEL MAPS OF CARBON SEQUESTRATION POTENTIAL; (5) WEB-BASED DATA VISUALIZATION AND QUERY SYSTEM; (6) MAP OF CANOPY STRUCTURE AND BIOMASS DERIVED FROM WALL-TO-WALL SINGLE PHOTON LIDAR FOR ALLEGHANY COUNTY;
Grant for Research
Contractor
University of Maryland (UNIVERSITY OF MARYLAND)
Contracting Agency/Office
National Aeronautics and Space Administration»Mission Support Directorate»NASA Shared Services Center
Effective date
09/12/2012
Obligated Amount
$1.3M
NNX12AM45H - SEMI-AUTONOMOUS TELEROBOTIC MANIPULATION FOR ON-ORBIT SPACECRAFT SERVICING AND ASSEMBLY OVER TIME-DELAYED TELEMETRY DESPITE THE PREVALENCE OF THIS TELEROBOTIC SURGERY OF HUMANS, WE STILL DO NOT HAVE THE CAPABILITY TO SERVICE (REFUEL AND REPAIR) OR ASSEMBLE SPACECRAFT ON-ORBIT WITH TELEROBOTIC SYSTEMS. WHETHER WE INTEND TO SERVICE ACTIVE SPACECRAFT ON-ORBIT, OR ASSEMBLE NEW LARGER SPACECRAFT ON-ORBIT, THE ABILITY TO DO SO WITH TELEROBOTIC SYSTEMS COULD YIELD A NEW PARADIGM FOR SPACECRAFT DEVELOPMENT AND DESIGN. THIS WOULD REMOVE THE CONSTRAINTS IMPOSED ON SPACECRAFT DESIGNS BY LAUNCH VEHICLES AND ENABLE THE DEVELOPMENT OF LARGER AND PREVIOUSLY-IMPOSSIBLE SPACE STRUCTURES. DEVELOPING SUFFICIENTLY CAPABLE AND ROBUST TELEROBOTIC SYSTEMS FOR ON-ORBIT OPERATIONS IS A TECHNOLOGICAL CHALLENGE THAT AND WILL REQUIRE NOVEL METHODS IN ORDER TO YIELD A PRACTICAL SOLUTION. MY PROPOSED RESEARCH WILL BEGIN WITH THE AUGMENTATION OF CLASSICAL TELEROBOTIC SYSTEMS WITH ALGORITHMS AND APPROACHES NORMALLY USED IN AUTONOMOUS ROBOTICS. I WILL DRAW ON MY OWN RESEARCH EXPERIENCE AND FAMILIARITY IN BOTH OF THESE SUBFIELDS OF ROBOTICS IN ORDER TO DEVELOP NEW FUNDAMENTAL RESEARCH IN SEMI-AUTONOMOUS TELEROBOTICS. MY RESEARCH PLAN INCLUDES: (1) INCORPORATING MORE SOPHISTICATED SENSING AND MOTION-PLANNING INTO TELEROBOTS, (2) DEVELOPING ALGORITHMS AND SYSTEMS FOR MULTI-SCALE AUTONOMY, AND (3) USING MACHINE-LEARNING TO FEED TELEROBOTIC EXPERIENCE BACK INTO THE SEMI-AUTONOMOUS SYSTEMS. I BELIEVE WE CAN DEVELOP A NEW CLASS OF ROBUST SEMI-AUTONOMOUS ROBOTIC PLATFORMS THAT WILL ENABLE US TO PERFORM FAR MORE COMPLEX TASKS THAN WE COULD PREVIOUSLY WITH TELEROBOTIC SYSTEMS UNDER HIGH-LATENCY TELEMETRY.
Training Grant
Contractor
Johns Hopkins (JOHNS HOPKINS UNIVERSITY, THE)
Contracting Agency/Office
National Aeronautics and Space Administration»Mission Support Directorate»NASA Shared Services Center
Effective date
08/07/2012
Obligated Amount
$211.1k
NNX12AN46H - DATA-DRIVEN GEOSPATIAL MODELS ARE WIDELY USED TO UNDERSTAND THE GLOBALLY CHANGING ENVIRONMENT, AND TO SUPPORT POLICY- AND DECISION-MAKING. CAREFUL MODEL EVALUATION IS ESSENTIAL FOR THESE PRACTICES. IT HAS TWO MAJOR GOALS: TO UNDERSTAND MODEL PERFORMANCE, AND TO PROVIDE GUIDANCE FOR MODEL IMPROVEMENT. CONVENTIONAL ERROR ANALYSIS METHODS PROVIDE LIMITED INSIGHTS ON THESE ISSUES. HENCE, I PROPOSE TO USE THE BIAS-VARIANCE (BV) ERROR DECOMPOSITION IN GEOSPATIAL MODELING. THIS APPROACH DECOMPOSES THE EXPECTED MODEL ERROR INTO BIAS, VARIANCE, AND NOISE. ORIGINATING IN STATISTICS AND MACHINE LEARNING, BV ANALYSIS HAS PROVEN USEFUL FOR UNDERSTANDING THE SOURCES OF MODELING ERROR, EXPLAINING MODEL PERFORMANCE, AND INFORMING MODEL IMPROVEMENT EFFORTS, BUT IT HAS NOT BEEN EXAMINED IN THE CONTEXT OF GEOSPATIAL MODELING. WITH THIS RESEARCH, I ATTEMPT TO MAKE THREE CONTRIBUTIONS. FIRST, I WILL DEFINE THE BV DECOMPOSITION FOR ABSOLUTE ERRORS. MODEL ERRORS CAN BE DEFINED IN DIFFERENT WAYS, SUCH AS SQUARED ERROR, ABSOLUTE ERROR, AND CATEGORICAL ERROR. ALL THREE ARE RELEVANT TO GEOSPATIAL MODELING, BUT EXISTING LITERATURE HAS NOT DEFINED THE BV DECOMPOSITION FOR ABSOLUTE ERRORS. SECONDLY, I WILL EXPLORE THE USEFULNESS OF EXAMINING DIFFERENT ERROR COMPONENTS SPATIALLY. I ANTICIPATE THE SPATIAL DISTRIBUTION ALONG WITH THE MATHEMATICAL MEANING OF BIAS AND VARIANCE WILL AID RESEARCHERS IN DEVELOPING EFFECTIVE MODEL IMPROVEMENT STRATEGIES. FOR INSTANCE, THE SPATIAL CLUSTERING OF BIAS AND VARIANCE MAY BE USED TO DETECT AND SURMOUNT SPATIAL NON-STATIONARITY, AND TO LOCATE ADDITIONAL TRAINING EXAMPLES THAT CAN IMPROVE ACCURACY. THIRDLY, SINCE GEOSPATIAL MODELS OFTEN USE SQUARED ERROR IN SITUATIONS WHERE ABSOLUTE OR CATEGORICAL ERROR IS THE NATURAL CHOICE, I WILL INVESTIGATE DIFFERENCES IN MODEL EVALUATION CONCLUSIONS THAT ARISE FROM THE CHOICE OF ERROR DEFINITION. IF OUR IMPRESSION ABOUT THE MODEL PERFORMANCE IS SENSITIVE TO THE WAY ERROR IS DEFINED, CARE IN CHOOSING AN ERROR MEASURE IS IMPERATIVE FOR APPROPRIATE MODEL EVALUATION.
Training Grant
Contractor
UNIVERSITY OF WISCONSIN SYSTEM
Contracting Agency/Office
National Aeronautics and Space Administration»Mission Support Directorate»NASA Shared Services Center
Effective date
08/07/2012
Obligated Amount
$30k
NNX12AE64G - CARNEGIE SCIENCE CENTER (CSC), LOCATED IN PITTSBURGH, PA, IS RECOGNIZED FOR ITS EXPERTISE IN TRANSLATING SCIENCE AND TECHNOLOGY INTO HANDS-ON LEARNING EXPERIENCES THAT EXCITE, EDUCATE, AND INSPIRE DIVERSE AUDIENCES. "SOLARQUEST: EXPLORATION OF THE SUN-EARTH SYSTEM" BUILDS ON THIS EXPERTISE, MARRYING CSC'S TRACK RECORD OF DEVELOPING AND DELIVERING EFFECTIVE SCHOOL OUTREACH PROGRAMMING WITH ITS LEADERSHIP IN ASTRONOMY AND SPACE SCIENCE EDUCATION THROUGH THE BUHL PLANETARIUM&OBSERVATORY (ESTABLISHED IN 1939) AND THE BUHL DIGITAL PLANETARIUM. "SOLARQUEST: EXPLORATION OF THE SUN-EARTH SYSTEM" WILL BE DEVELOPED FOR CSC'S SCIENCE ON THE ROAD PROGRAM, WHICH CURRENTLY DELIVERS STANDARDS-BASED EDUCATIONAL EXPERIENCES TO 200,000 K-8 STUDENTS AND TEACHERS ANNUALLY IN 150 LOCATIONS ACROSS 5 STATES. THE PRIMARY PREMISE OF THE SOLARQUEST PROJECT IS TO PROVIDE STUDENTS IN GRADES 4-8 WITH HIGH-QUALITY, HIGH-TECHNOLOGY LEARNING EXPERIENCES IN HELIOPHYSICS THAT WOULD NOT OTHERWISE BE ACCESSIBLE TO THEM. SOLARQUEST WILL INCORPORATE FORMATS THAT HAVE PROVEN TO BE SUCCESSFUL IN ENGAGING YOUNG MINDS AND IMAGINATIONS, UTILIZING NASA'S SOLAR DYNAMICS OBSERVATORY (SDO) MISSION DATA, IMAGERY, AND EXPERTS TO PORTRAY THE ENVIRONMENT IN WHICH THE EARTH TRAVELS AND THE WAYS IN WHICH THE SUN AND ITS INTERACTIONS WITH THE PLANET AFFECT HUMAN LIFE. THE PRIMARY COMPONENTS OF THE PROJECT INCLUDE A 45-MINUTE, IMMERSIVE ASSEMBLY PROGRAM THAT INCORPORATES A 10-MINUTE DIGITAL PLANETARIUM SHOW, AND HANDS-ON ACTIVITIES THAT REINFORCE THE SCIENTIFIC CONCEPTS PRESENTED IN THE ASSEMBLY SHOW. TEACHER RESOURCES WILL BE PROVIDED ON THE CSC WEB SITE. ALL PROGRAM CONTENT AND ACTIVITIES WILL BE AGE/GRADE ADAPTED AND ALIGNED WITH NATIONAL SCIENCE EDUCATION STANDARDS; THE MOST IN-DEPTH CONNECTIONS WILL BE WITH CONTENT STANDARDS B: PHYSICAL SCIENCE; D: EARTH AND SPACE SCIENCE; AND E: SCIENCE AND TECHNOLOGY. LIKE OTHER SCIENCE ON THE ROAD ASSEMBLY SHOWS (E.G., "ION JONES AND THE LOST CASTLE OF CHEMISTRY" AND "CAPTAIN GREEN'S TIME MACHINE"), SOLARQUEST IS CONCEIVED AS AN ADVENTURE STORY DELIVERED THROUGH A LIVELY MIX OF LARGE-SCALE PROPS AND COSTUMES, CHEMISTRY AND PHYSICS DEMONSTRATIONS, VIDEO, AND AUDIENCE PARTICIPATION. ALL ASSEMBLY SHOWS ARE LED BY AN EXPERIENCED CSC EDUCATOR/PRESENTER. IN THE CASE OF SOLARQUEST, STUDENTS WILL BOARD AN IMAGINARY SPACECRAFT, LED BY A CSC "COMMANDER," AND BLAST OFF TO EXPLORE THE SUN AND ITS IMPACT ON PLANET EARTH. ALONG THE WAY, THEY WILL VISIT (VIA VIDEOTAPE) NASA SATELLITES ENGAGED IN HELIOPHYSICS RESEARCH AND ENCOUNTER NASA SDO SCIENTISTS AND EXPERTS WHO DISCUSS THEIR WORK AND THEIR CAREERS. THE PROGRAM WILL BE ORGANIZED AROUND MAJOR THEMES, SUCH AS ULTRA-VIOLET LIGHT, SOUND, COLOR, SATELLITE DESIGN, AND SPACE WEATHER, AMONG OTHERS. THESE THEMES, AND CORRESPONDING SCIENTIFIC CONCEPTS, ARE REINFORCED IN 10 HANDS-ON ACTIVITY STATIONS THAT STUDENTS WILL EXPERIENCE FOLLOWING THE ASSEMBLY SHOW. THE 10-MINUTE DIGITAL PLANETARIUM SHOW, DELIVERED TO SCHOOL AUDIENCES VIA A PORTABLE DIGITAL PLANETARIUM SYSTEM, BRINGS HELIOPHYSICS TO LIFE IN A UNIQUE AND POWERFUL WAY--A STUNNING "YOU ARE THERE" PERSPECTIVE THAT IS NOT POSSIBLE THROUGH OTHER PRESENTATION FORMATS. AUDIENCES WILL LITERALLY "FLY THROUGH" THE SUN-EARTH ENVIRONMENT, EXPERIENCING THE SUN-EARTH SYSTEM THROUGH THE LENS OF ACTUAL NASA SDO MISSION DATA. THE DIGITAL PLANETARIUM SHOW WILL HAVE A SECOND OUTLET AS PART OF CSC'S REGULAR PROGRAMMING IN THE BUHL DIGITAL PLANETARIUM. CSC ALSO PROPOSES TO DISTRIBUTE 50 FULL-DOME DIGITAL COPIES OF THE SHOW TO DIGITAL-CAPABLE PLANETARIUMS AROUND THE WORLD. "SOLARQUEST: EXPLORATION OF THE SUN-EARTH SYSTEM" IS PROJECTED TO HAVE A 5-YEAR LIFE SPAN FOLLOWING ITS FULL LAUNCH IN 2013. APPROXIMATELY 850,000 YOUNG PEOPLE AND ADULTS COULD BE SERVED BY THE PROJECT: 250,000 STUDENTS AND TEACHERS THROUGH SCIENCE ON THE ROAD; 100,000 THROUGH PRESENTATION AT THE BUHL DIGITAL PLANETARIUM; AND 500,000 AT PLANETARIUMS WORLDWIDE.
Grant for Research
Contractor
CARNEGIE INSTITUTE
Contracting Agency/Office
National Aeronautics and Space Administration»Mission Support Directorate»NASA Shared Services Center
Effective date
06/18/2012
Obligated Amount
$764.4k
NNX11AQ55G - THE MAGNETIC STRUCTURE AND ITS EVOLUTION IN ACTIVE REGIONS ARE DETERMINING FACTORS OF SOLAR ERUPTIONS SUCH AS FLARES AND CORONAL MASS EJECTIONS (CMES). USING DATA FROM SOLAR DYNAMIC OBSERVATORY (SDO) ALONG WITH OTHER SPACE AND GROUND-BASED OBSERVATIONS, WE PROPOSE TO STUDY THE RELATION BETWEEN EVOLUTION OF SOLAR MAGNETIC FIELDS AND FLARES/CMES, AND FURTHERMORE ESTABLISH A SYSTEM OF FLARE/CME PREDICTION THAT INVOLVES TWO LINKED COMPONENTS: (1) WE WILL ESTABLISH STATISTICAL CORRELATION BETWEEN MAGNETIC FIELDS MEASURED IN SOLAR SURFACE AND PRODUCTIVITY OF FLARES/CMES. IN THE PAST, WE HAVE EXTENSIVELY STUDIED THE FOLLOWING PARAMETERS: UNSIGNED MAGNETIC FLUX, MAGNETIC GRADIENT ALONG THE NEUTRAL LINE AND MAGNETIC ENERGY DISSIPATION. IN THE PROPOSED STUDY WE WILL EXPANDING THE PARAMETERS TO INCLUDE (A) MAGNETIC HELICITY INJECTION (BASED ON TRACKING THE MOTION OF SURFACE MAGNETIC FIELDS) WHICH TAKES THE EVOLUTION OF MAGNETIC FIELDS INTO CONSIDERATION, (B) PHOTOSPHERIC EXCESS ENERGY AND (C) FREE MAGNETIC ENERGY WHICH WOULD BE THE MOST DIRECT PARAMETER DESCRIBING THE AVAILABLE ENERGY IN SOLAR CORONA TO POWER FLARES/CMES. THE FIRST TWO PARAMETERS CAN BE DERIVED FROM PHOTOSPHERIC VECTOR MAGNETOGRAMS WHILE THE THIRD ONE REQUIRES EXTRAPOLATING THREE-DIMENSIONAL (3-D) NON-LINEAR FORCE-FREE (NLFF) FIELDS USING THE OBSERVED VECTOR MAGNETOGRAMS AS THE BOUNDARY CONDITION. WE ARE CAUTIOUS THAT THE PROPOSED WORK WILL NOT LARGELY RELY ON THE NLFF EXTRAPOLATION. THE FIRST TWO PARAMETERS CAN BE STUDIED WITH CONFIDENCE. WE WILL USE SDO/AIA DATA FOR VALIDATION OF NLFF EXTRAPOLATION AND THE USE OF THE THIRD PARAMETER. IN ADDITION TO THE CORRELATION STUDY, USING THE SAME DATA SETS TO FOLLOW THE EVOLUTION OF MAGNETIC FIELDS IN FLARE PRODUCTIVE REGIONS WILL LIKELY DISCLOSE THE TRIGGERING MECHANISM OF ERUPTIVE FLARES/CMES, SUCH AS FLUX EMERGENCE, FORMATION OF MAGNETIC CHANNEL AND INJECTION OF HELICITY IN THE OPPOSITE SIGN. (2) USING ESTABLISHED CORRELATION BETWEEN MAGNETIC PARAMETERS AND THE FLARE PRODUCTIVITY, WE WILL DEVELOP STATISTICAL AND MACHINE LEARNING TECHNIQUES TO PREDICT OCCURRENCE OF SOLAR FLARES. BASED ON PRELIMINARY RESULTS OBTAINED SO FAR, A PROMISING TOOL THAT WE WILL DEVELOP IS A COMBINATION OF ORDINAL LOGISTICAL REGRESSION AND SUPPORT VECTOR MACHINE (SVM) CLASSIFICATION. ALTHOUGH OUR CODES ARE DEVELOPED FOR FLARE FORECASTING, THEY CAN BE ADAPTED FOR CME FORECASTING. WE WILL EXPLORE THE RELATION BETWEEN MAGNETIC HELICITY AND CME OCCURRENCE. OUR PROPOSED RESEARCH ALIGNS WITH THE LIVING WITH A STAR (LWS) STRATEGIC GOALS NO.1 IN THAT IT HELPS TO ``DELIVER THE UNDERSTANDING AND MODELING REQUIRED FOR USEFUL PREDICTION OF THE VARIABLE SOLAR PARTICULATE AND RADIO-ACTIVE ENVIRONMENT". THE PROPOSED RESEARCH ADDRESSES THE ``SCIENCE ANALYSIS FOR THE SDO" COMPONENT OF THE LWS TARGETED RESEARCH AND TECHNOLOGY PROGRAM (TR&T), IN PARTICULAR, THE ``USE OF OBSERVATIONS TO PREDICT FUTURE SOLAR ACTIVITY". OUR RESEARCH WILL DELIVER NEW UNDERSTANDING OF SOLAR ERUPTIONS THROUGH THE ENERGY AND HELICITY BUDGETS OF ACTIVE REGIONS, AND WILL THEREFORE CONTRIBUTE TO PREDICT FLARES/CMES.
Grant for Research
Contractor
New Jersey Institute Of Technology (NEW JERSEY INSTITUTE OF TECHNOLOGY)
Contracting Agency/Office
National Aeronautics and Space Administration»Mission Support Directorate»NASA Shared Services Center
Effective date
09/12/2011
Obligated Amount
$309k
NNX09AK60G - CORNELL UNIVERSITY, INC. QUANTIFIED UNCERTAINTY: FLEXIBLE PROBABILISTIC MODELING OF DYNAMIC SPECTRA AND OTHER ASTRONOMICAL DATA MODERN ASTRONOMICAL INSTRUMENTS PROVIDE DATA OF INCREASING QUANTITY AND DETAIL, STRAINING THE CAPABILITY OF CONVENTIONAL ANALYSIS TECHNIQUES THAT RELY ON INFLEXIBLE PARAMETERIZED MODELS. RELYING ON RECENT DEVELOPMENTS IN STATISTICS AND MACHINE LEARNING, WE WILL BUILD FLEXIBLE, ADAPTIVE MODELS FOR ASTROPHYSICAL SOURCES WITH COMPLEX, MULTI-BAND LIGHT CURVES (DYNAMIC SPECTRA). TWO COMPELLING ASTRONOMICAL PROBLEMS WILL BE THE PRIMARY APPLICATIONS: (1) ANALYSIS OF SPECTRO-TEMPORAL PROPERTIES OF PULSES COMPRISING GAMMA-RAY BURSTS (GRBS), WITH GOALS OF CONSTRAINING CENTRAL ENGINE MODELS AND UNCOVERING CORRELATIONS THAT WILL IMPROVE OUR CAPABILITY TO USE GRBS AS PROBES OF THE HIGH-REDSHIFT UNIVERSE. (2) ANALYSIS OF MULTICOLOR LIGHT CURVES OF TYPE IA SUPERNOVAE (SN IA), WITH GOALS OF IMPROVING THEIR CALIBRATION AS STANDARD CANDLES FOR MEASURING PROPERTIES OF THE DARK ENERGY CAUSING THE ACCELERATED EXPANSION OF THE UNIVERSE. A SECONDARY APPLICATION AREA IS MODELING LIGHT CURVES FROM ACTIVE GALACTIC NUCLEI (AGN), INCLUDING GRAVITATIONALLY LENSED AGN AND MULTIWAVELENGTH AGN DATA. WE WILL USE GAUSSIAN PROCESS AND LEVY ADAPTIVE REGRESSION KERNEL METHODS TO BUILD FULLY PROBABILISTIC MODELS FOR THESE AND OTHER COMPLEX SIGNALS, ENABLING FULL QUANTIFICATION OF UNCERTAINTIES, INCLUDING UNCERTAINTIES IN THE NUMBERS OF COMPONENTS COMPRISING THE SIGNALS. THIS WORK WILL IMPROVE THE SCIENCE RETURN FROM NASA'S GRB MISSIONS, AND PROVIDE KEY ENABLING TECHNOLOGY FOR THE UPCOMING JDEM MISSION THAT WILL USE SN IA TO MEASURE DARK ENERGY.
Grant for Research
Contractor
CORNELL UNIVERSITY, INC
Contracting Agency/Office
National Aeronautics and Space Administration»Mission Support Directorate»NASA Shared Services Center
Effective date
07/17/2009
Obligated Amount
$674.4k

Let's get started today

Let's Get Started Today

Contracting Resources

Awarded Contracts within Office of the Administrator

Let's get started today

Let's Get Started Today

Awarded Contracts by Industry

Let's get started today

Let's Get Started Today