ORNL Knowledge Discovery Science Agenda

1
ORNL Knowledge Discovery Science Agenda
Brian Worley Director Computational Sciences
and Engineering Division
2
ORNL Is Committed to the Knowledge Discovery
Agenda

• Entire Research Division Focused on Knowledge
  Discovery
• Appropriate Resources
• HPC, Networking, MRF, JICS
• LDRD Initiative in Knowledge Discovery
• Programmatic efforts well-aligned with this
  science agenda

3
ORNL Focus in Knowledge Discovery

• Actionable insights from massive, dynamic,
  disparate data sources
• Ability to ask more complex questions and detect
  more complex processes using increasingly higher
  data resolution

Dynamic Disparate Sensor Data Massive Simulated
Data Data Driven Simulations Hypotheses
Generation Heightened Sense of Urgency
Actionable Knowledge Decision Support
Discovered Data Experimental Data Pattern
Detection Infer Physical Law Model Development
Model Validation New Science New Technology
4
Ubiquitous Data
Reality
Amounts of analyzable data
New paradigms
Data collected
No paradigm change
Data analyzed
50s
60s
70s
80s
90s
21st Century ?
Time
5
Knowledge Management and Knowledge Discovery will
become more Integrated Disciplines
KD E.g., Traditional Data-Mining
KD E.g., Mash-ups, Distributed Data-mining
KD E.g., Complex Event Processing, multi-dimensio
nal/multi-modal stream correlation
KM E.g., RFID streaming applications, distributed
federated servers
KM E.g., Streaming video servers, web-services
KM E.g., Traditional Databases
Percent of Data that is Real-Time
6
Information Extraction and Fusion

• High-speed document clustering
• Advanced image and text search
• High-speed information fusion
• Content extraction
• Name disambiguation
• Place disambiguation
• Deception detection

7
Real-Time Knowledge Discovery
- Provide decision support in minutes - Provide
real time interview support

• Detect anomalies
• Data dip into structured and unstructured data
• Hypothesis generation
• Complex event processing
• Threat anticipation

8
ORNL SensorNet Program

• Interdiction, detection, emergency response
• Mobile, Transportation Corridors, Ports, Military
  Bases
• Real-Time Data Management
• Collection, Dissemination, Archiving
• Pre-deployment analysis
• Cost, Performance Prediction, Risk vs Benefit
• Wide-area ubiquitous sensing, actuation, and
  deployment
• Orchestrating the functionality across a large
  system of distributed sensors/processors (eg
  Electric Grid, Autonomous robotic systems)
• Cross-agency and cross-administrative boundary
  data-sharing and interoperability
• Standards and policies
• Net-Centric Services
• Security, Access Controls

9
SensorNet Interoperable Net-Centric Data
Architecture

• Scalable
• Standards-based
• Interoperable
• Seamless
• Secure

911 CAD
Database
External LEA and DoD Databases
Mass Notification System
911 Communications Server
(OGC Sensor Alert Service)
(OGC Web Feature Services)
(IEEE 1451 proxy services)
Net-Centric Enterprise Architecture
CheckPoint Duress and Intrusion Alarms
Honeywell and Monaco Fire Alarms
ObjectVideo Intelligent Video Surveillance
Lowery/CCIS Access Control
NOAA and Smiths Sensors
Nextel Tracking Service
10
Electric Grid Situational Awareness and Analysis
Live feeds (w/TVA and other utilities), Weather,
Hurricane Trajectories and other contingencies
and analysis
11
Real-Time Data-Driven Simulation - Scaling to
Finest Granularity of Information
Emergent behavior
Sensor networks
High-performance computing
12
Achieving Systematic Situation Awareness through
Empirical Syndromic Surveillance

• Collect and evaluate dynamic data from various
  sources
• Integrate data over long periods of time and
  build knowledge base
• Detect current potential anomalous events
• Disseminate pertinent information and tasking to
  evaluators layered in a rational hierarchy
• Collect vetting of anomaly and proposed actions
  among evaluators at appropriate levels in a
  timely fashion
• Cross check and share evaluation and recommended
  actions between evaluators
• Update knowledge base with vetting results and
  proposed actions

13
Quantum Information Systems - A
New Focus at ORNL

• QIS is a national priority
• Goals for ACI Research Overcoming the
  technological barriers to the practical use of
  quantum information processing to revolutionize
  fields of secure communications, as well as
  quantum mechanics simulations used in physics,
  chemistry, biology, and materials science (DoE,
  NIST, NSF)
• --American Competitiveness Initiative,
  February, 2006.
• Domestic Policy Council, Office of Science and
  Technology Policy.
• ORNL is investing in facilities for QIS
• ORNL is hiring new staff dedicated to QIS
• ORNL is making internal research investments in
  QIS

14
CSED Core Research Areas

• Modeling and Simulation
• Physics-based predictive simulations
• Parallel discrete event simulations
• Information Security
• Information assurance
• Quantum Information Systems

• Information Systems
• Systems architecture and design
• Large scale data management
• Geospatial Sciences
• Population and social dynamics
• Feature and process extraction
• Information Analysis
• Agent-based methods
• Text analysis

15
CSED Current Main RD Activities

• Modeling and Simulation
• Sensor based assessment and mission support
• Discrete event simulations
• Behavioral sciences
• Complex nonlinear systems
• Biomedical applications
• Power grid simulation and control
• Physics based modeling

• Information Systems
• Systems architecture and design
• Large scale data management and integration
• Sensor data assimilation
• Logistics and asset visibility
• Integrated emergency response
• Risk assessment and other decision tools

16
CSED Current Main RD Activities (cont)

• Geospatial Sciences
• Geographic Data Sciences
• High-performance visualization
• Geocomputation for Transportation
• Population and social dynamics
• Emergency response and resilience
• Advanced Geospatial Applications

• Information Analysis
• Relation of people, places, and events in large
  document sets
• Anomaly detection in military transportation data
• Geospatial agents
• Swarming methods
• Content extraction in video streams
• Advanced computing for information analysis

17
CSED Current Main RD Activities (cont)

• Information Security
• Insider threat detection mitigation
• Information Operations
• Information flow accountability and infrastructure
  trustworthiness
• Applied information and decision theory
• Waveguide entangle photon source
• Cultural behavioral assessments