Computer Science & Informatics Dept.

CIS Informatics Research Strategic Plan

Informatics is the science that encompasses powerful data methods and relationship-based analytic methods to support decision-making, and decision-makers, when data is so complex or so voluminous as to defy effective use for detailed understanding.

Informatics encompasses complex information-based analytical capabilities to support decisions in uncertain environments. The area of informatics has many technical and science challenges:

Scalability – algorithms and architectures
Complexity – “the big data problem” and/or complex relationships
Uncertainty – a driving concern, with “trust” issues for decisions
Validation – a different paradigm from scientific computing
Prediction – inference of relationships, using statistics or computing methods

We have the opportunity to leverage decades of experience in High Performance Computing for science and physics-based applications. This expertise can be leveraged and refactored for informatics challenges critical to the Nation.

Vision:
To differentiate ourselves as the informatics and scalable analysis research organization for decision-support in national security and national-level challenges.

Mission:
CIS will serve as the internal research organization to Sandia mission SMUs. Our purpose is to build foundational capabilities, in support of the Nation, through informatics research to address “the big data problem” that is prevalent in numerous Sandia mission challenge areas. Our ability to partner and carry research to impact mission applications will be foundational for Sandia mission challenges in cyber, all things nuclear (nonproliferation and the terrorism threat), and energy, including our longstanding mission in defense.

Strategic Objectives:

Since Informatics is a relatively new research area and has no S&T strategic plan in place, this planning document necessarily begins with the basics: establish a plan to plan the informatics R&D, establish partnerships and collaborations, and identify differentiating research investments.

Objective: 1.0 Informatics Research Program and Development Plan
Build a sustainable CIS S&T informatics research program that identifies and develops differentiating strengths in our science and technology capabilities, addresses capability gaps and research opportunities, and leverages funding streams from multiple research funding sources.
Timeframe: develop immediately, maintain guiding plan as a living document

Goal: 1.1 Differentiating Strengths
Identify and build on differentiating strengths in informatics science and technology with the potential to impact national security and other national-scale challenges in energy and defense.

Milestone: 1.1.1 Informatics Roadmap
Develop a roadmap for informatics that provides an overall plan for the research and shows how the various parts (and funding streams) integrate into a strategic whole. Note that the roadmap should be a “living document.”
Timeframe: immediate and ongoing

Milestone 1.1.2 Opportunity/influence strategy
Develop a plan to influence and to respond to opportunities (e.g., BAAs, proposal calls, etc) from research funding sources, while building strategic capability that follows the informatics roadmap.
Timeframe: develop immediately

Milestone 1.13 Investment/disinvestment strategy
Develop a CIS evaluation strategy for investment/maintenance/disinvestment of research areas and infrastructure.

Goal: 1.2 Research and Development to Target Informatics and
Capability Gaps
Identify capability opportunities to target in the CIS informatics research program.
Timeframe: 1-3 year objectives that can be addressed in 3-year research projects.

Milestone: 1.2.1 Internal R&D focus areas
Work with LDRD Investment Area Teams (IATs) on more-focused and strategic LDRD Calls with strategic management participation on IATs. Work with DOE/ASC program on informatics challenges appropriate for FRIC, CSSE, and CS research foundation funding.
Timeframe: New LDRD process implemented for FY10

Milestone: 1.2.2 External R&D challenges and opportunities
Work with existing sponsors and funders (e.g., in DOE, Office of Science, NNSA, DoD, intelligence community, DARPA, ARPA-E, etc) to identify and respond to national-scale informatics challenges and national lab opportunities.

Goal: 1.3 Self-sustaining Research
Establish a strategy to sustain longer-term research for mission and national needs.

Milestone: 1.3.1 Enduring external research relationships
Establish enduring relationships with traditional (to CIS) and non-traditional external research organizations.

Milestone: 1.3.2 National technical service and influence
Participate on national workshops, forums, and committees to influence the future research directions of CIS-traditional and non-traditional national research organizations. More closely align research with strategic national security needs and national-scale challenges critical to the Nation’s future.

Goal: 1.4 Resource Investment Plan for Research
Establish a strategy to best invest and focus our resources for research, mission, and national needs.

Milestone 1.4.1 People
Hire, matrix, partner, nurture, and retain. Plan for customer support and software/hardware maintenance of capabilities (e.g., through internal support/partnerships, contracts, etc).

Milestone 1.4.2 Computing Infrastructure
Plan for current and future computing infrastructure and capability support. Includes hardware/software support, funding mechanisms, programming models tailored to the architecture, evaluation of the architecture using test suite of applications.

Milestone 1.4.3 Approaches for intellectual property
Make a priori decisions on intellectual property (e.g., patent, copyright, trademark, open source release to set standards, limited release, etc). Consider the impact of partnerships and legal arrangements on intellectual property (e.g., CRADAs, consortiums, WFO work, etc).

Objective: 2.0 Informatics Partnerships and Collaborations
Foster partnerships, internally with SNL SMUs and externally with academia and industry, that enable strategic collaborations between informatics research and more focused applications, addressing national security and national-scale challenges with mission-focus, including: national security, cyber, defense, energy, economic security, and health care challenges for the Nation.
Timeframe: 1-3 years

Goal: 2.1 Strategic Partnerships and Collaborations with Mission-
focused Organizations
Establish and increase partnerships and collaborations between CIS and SMUs (DS&A, ERN, HSD, DP). Leverage our differentiating strengths through CIS partnerships with SMUs to collaboratively provide uniquely new capabilities to sustain and grow existing SMU sponsors and customers and to help engage new sponsors and customers.

Milestone: 2.1.1 Business models and sponsor/customer engagement models:
Develop business models and collaboration plans between CIS and SMUs. Include process and plans for engagement with external sponsors/customers (e.g., CIS directly engage sponsor, CIS jointly engage sponsor, CIS serve as research resource for another SNL organization’s engagement). Identify point-of-contact/“owner” for external engagements.

Milestone: 2.1.2 (I)WFO model with applications’ customers and sponsors to sustain research:
Develop a model for engagement of shorter-term and longer-term work-for-others that sustains longer-term research and computing infrastructure. Includes forming strategic partnerships where CIS participates in initial sponsor/customer engagement.

Goal 2.2 Internal Research Collaborations
Strengthen research collaborations within CIS between 1400 and 8900, and with 6300, 9500 and other research organizations across Sandia.

Milestone: 2.2.1 New model to sustain research
Participate in developing new Sandia Laboratories’ models that sustain internally-funded research and make Sandia a national resource for research and mission-focus areas.

Goal 2.3 Research Collaborations with Other Labs
Build and strengthen research collaborations with other laboratories. In an increasingly competitive environment, building on our strengths and collaborating to leverage others’ strengths will best serve the Nation.

Milestone: 2.3.1 Inter-Lab collaborations
Establish opportunities to collaborate, rather than compete, with other Labs, including energy, defense, and science labs.

Goal: 2.4 External Collaborations with Academia and Industry
Build and strengthen research collaborations with universities, industries, and the private sector.

Milestone: 2.4.1 Academic and industrial collaborations
Collaborate with academic institutions and industry to avoid replicating existing capability. Utilize unique facilities (e.g., CSRI), contracts, partnerships, and professional interactions.

Objective 3.0 Informatics Science and Technology Research Investment Areas
Establish a shared CIS technical research vision for informatics to guide management and staff in investment decisions, research goals, and cross-cutting milestones.

Goal 3.1 Analysis Algorithm Thrusts
Build a leadership position (where CIS is known for the science) in innovative algorithms that are scalable and/or highly sophisticated. Grow areas where CIS can leverage existing capabilities.

Milestone: 3.1.1 Leadership in analysis algorithms
Establish a CIS leadership position by building new array-based and graph-based informatics capabilities. Algorithm areas include multi-dimensional analysis, tensors, algebraic methods, and graph analytics applicable to data such as text, networks, time series, and images. Partner with researchers from academia, industry, external agencies, and Sandia-internal organizations.

Milestone: 3.1.2 Leadership in the area of data analytics
Establish a leadership position in data analytics by leveraging existing intellectual expertise and capabilities (e.g., classification analysis, linear and nonlinear solvers, machine learning, etc) for challenges in discrete-based uncertainty quantification, sensitivity analysis, and data/network characterization. Research and develop statistics-based methodologies for uncertainty and prediction in informatics, incorporating analyst concerns regarding “trust” and uncertainty in life-critical assessments. Partner with researchers from academia, industry, external agencies, and Sandia-internal organizations.

Milestone: 3.1.3 Growth in scalable algorithms
Grow existing CIS capabilities by scaling algorithms to ever-larger “big data problems” and by parallelizing (serial) analysis algorithms. Example areas include large-scale analysis, information visualization, feature extraction, data analysis methods, dimensionality reduction, and simulation.

Milestone: 3.1.4 New methods for stream processing
Develop real-time data analyses, methodologies, and/or heuristics for streaming data applications (such as cyber security and military systems) to leverage existing research methods for new challenges (e.g., real-time anomaly detection and response, battlefield communications, where numerous data streams dominate).

Goal: 3.2 Computer Architecture Thrusts
Build a leadership position (where CIS is known for the science) in processor architectures for informatics. Evaluate and better understand the capabilities of various architectures.

Milestone: 3.2.1 Leadership in processor architectures
Build a leadership position in research and development of new processor architectures for informatics applications (e.g., unstructured data applications with minimal computation, maximal data fetch, typically unpartitionable, out-of-core datasets) where data access patterns are relatively random and fine-grained.

Milestone: 3.2.2 Performance evaluation of integrated data systems architecture
Implement and test performance optimizations for data movement and processing within an HPC informatics system architecture which utilizes an effective mix of architectures. These architectures to be integrated may include: data servers/systems, data storage/warehousing appliances, data management (representation, storage, and retrieval), informatics analysis machines, distributed-memory machines, and user-interface/delivery approaches (e.g., client-server for visualization).

Milestone: 3.2.3 Methodologies for performance evaluation
of architectures
Develop methodologies to evaluate the performance of different computer architectures on a common test problem suite (including clustering, unstructured graph relationships, scientific computations) to better understand the favored usage for various architectures.

Goal: 3.3 Human-Computer Interaction Thrusts
Build a leadership position (where CIS is known for the science) in human-computer interaction areas, including information visualization and human factors. Grow integration areas where CIS can leverage analysis algorithms and provide the capability for interoperability among algorithms.

Milestone: 3.3.1 Leadership in information visualization
Build a leadership position for highly-scalable information visualization capabilities to visually analyze and adaptively display analysis outputs for user understanding.

Milestone: 3.3.2 Leadership in human factors
Establish a leadership position in human factors to elicit requirements from analysts, to bridge the gap between analysts and researchers, and to articulate needs to guide research. Addresses user adoption challenges and usage issues (e.g., use cases to define analyst processes, human-computer interface, usability, feedback, and metrics).

Milestone: 3.3.3 Integration for informatics framework
Develop an adaptive integration framework for CIS capabilities (e.g. our Titan framework) to give a common user interaction framework for informatics analysis algorithms that provides the capability for interoperability across algorithms.