KDI Initiative provides 48 Million in Grants

By Louise Arnheim

Within days of receiving the final go-ahead from the House, the National Science Foundation (NSF) released a $48 million proposal solicitation for its Knowledge and Distributed Intelligence (KDI) initiative. KDI is an NSF-wide effort to capitalize upon recent achievements in computing power and connectivity.

Over the next few months, the Foundation will be seeking research proposals in three multidisciplinary program areas: Learning and Intelligent Systems (LIS), Knowledge Networking (KN), and New Computational Challenges (NCC). Individuals and groups from US colleges, universities, and nonprofit research institutions may apply for the grants (and NSF notes that industry partnerships are "encouraged"). Proposals are due May 8, with letters of intent expected by April 1.

Mike McCloskey, Program Manager, Human Cognition and Perception, is Chair of NSF’s internal working group on KDI, which meets weekly and is comprised of program officers from the six NSF directorates: Biological Sciences (BIO); Computer and Information Science and Engineering (CISE); Education and Human Resources (EHR); Engineering (ENG); Mathematical and Physical Sciences (MPS); and Social, Behavioral, and Economic Sciences (SBE). The group will manage the competition, solicitation, and review of submitted grants. It will also develop a plan for KDI’s future.

McCloskey, in a recent talk with CRN regarding the proposal solicitation, stated, "We’re now at a point where there have been huge advances in computer power and computer connectivity," he said. "The aim of KDI is to do everything we can to maximize the extent to which we can exploit these developments."

In several ways, KDI affirms the adage that "everything old is new again." First, KDI is the latest in a series of Foundation activities such as the new supercomputing partnerships (Partnerships for Advanced Computational Infrastructure), or PACI (see CRN, May 1997) and the Digital Libraries Initiative. Over $355 million has already been allocated for these ongoing projects.

Second, New Computational Challenges (NCC) is a continuation of NSF projects such as the Grand Challenge.

Third, another existing program, Collaborative Research on Learning Technologies (CRLT), was recently made a part of LIS. In fact, a first solicitation for LIS was issued in late 1996 (the announcement followed a series of NSF-sponsored workshops held between the fall of 1995 and spring of 1996). Then, in October 1997, NSF announced $22.5 million in grants for these 28 projects.

Also under the KDI umbrella is the Foundation’s part in the Next Generation Internet (NGI) initiative [NSF is one of several federal agencies such as the Defense Advanced Research Projects Agency (DARPA) and the National Aeronautics and Space Administration (NASA) that are involved in NGI-related research (see CRN, November 1998; also September 1997)]. Congress set aside a separate $10 million under KDI for the Foundation’s efforts at "enhancing Internet capabilities," but no grants will be awarded in this area.

Though NSF officials testified before Congress last year on KDI and funding was approved for FY ’98, federal lawmakers wanted more specific information. Quoting the conferees’ report: "The Foundation has not yet provided appropriate milestones or guideposts, to be accomplished in fiscal year 1998, and against which the agency can be measured in determining funding for fiscal year 1999. The conferees expect to receive references on such milestones and guideposts before the Foundation obligates any further funding for these programmatic areas." NSF, therefore, had to take the extra step of providing information before it could release the solicitation. Late in January, NSF received approval from the House and released the solicitation February 2.

NSF wants to see similar work emerge from the KDI grants as did with the LIS grants, so last year’s awards are a useful comparison point for prospective applicants. Among the largest (institutional) grantees were: Carnegie Mellon University, awarded $3,242,900; Brown University, awarded $2,323,400; Northwestern University, awarded $2,156,075; and Johns Hopkins University, awarded $1,548,900. Grant projects included statistical reasoning, visualization and the analysis of large-scale multiple-media data, computation in the visual domain, and reflective inquiry in knowledge-rich environments. (For a complete list of awards, projects, and institutions, see http://www.ehr.nsf.gov/lis/award97.htm.)

FY ‘98 Solicitation

As many as 75 projects could receive funding under KDI this year. The maximum amount available for any one project is $1 million annually for three years.

In its announcement, NSF stresses that it is looking for research that is "inherently multidisciplinary, or that, while lying within a single discipline, has clear impact on at least one other discipline." Below is an overview of the work NSF wants to see (to view the KDI announcement, including information regarding submission requirements, see http://www.nsf.gov/pubs/1998/nsf9855/nsf9855.htm)

Learning and Intelligent Systems – Based on the progress to date in cognitive science, computer science, and learning technologies, said McCloskey, "we are now ready to really move forward in our understanding of how the brain works, in particular, how people learn, and what constitutes intelligence; not only in living systems, but also in engineered systems." The solicitation describes three "research emphases" which are, briefly: (1) the identification, analysis, and modeling of natural and artificial systems; (2) the development of learning tools and methods that utilize various concepts (e.g., behavioral, cognitive) with different technologies (interactive, collaborative, multisensory); and (3) the stimulation of basic research in learning and intelligence with the goal of applying the results to numerous situations.

Knowledge Networking – This area, said McCloskey, covers activities such as remote data collection, remote control instrumentation, distributed research groups in remote locations, as well as new techniques for data mining.

In the solicitation, several categories and subcategories of research are identified:

1. Foundational Research

· Processes and dynamics of distributed intelligence

· Managing heterogeneity and achieving interoperability

· Computational infrastructure, tools and environments

2. Prototype Development and Research

3. Ethical, Social and Behavioral Research

· Knowledge dissemination and sustainable use of knowledge networks

· Social integration and impacts of knowledge networking

New Computational Challenges – For this category, NSF defines two areas of study: (1) problems of scale and structure; and (2) interplay between computations and data.

The idea here, said McCloskey, "is that throughout science and engineering, there are a lot of problems that involve extremely complex computation, and, in fact, computational problems that were intractable until recently." Having both the raw computer power as well as some new computational methods will enable researchers to study complex interactions such as neuron activity in the brain.

One final question asked of McCloskey was whether there is specific timeline (e.g., five years, ten years) for KDI? "This is a major theme for the Foundation right now," says McCloskey. "Certainly it’s expected to continue for a number of years."

NSF expects to issue another KDI solicitation in the near future with a proposal due date of February 1, 1999.