Dear Colleague, The area of pattern discovery has become critical both because of the scientific challenges involved in dealing with massive amounts of data and because of its relevance to national security needs. While steady incremental progress based on classical paradigms have made solutions to a few specific problems technologically feasible, for more complex tasks there is a clear need and scope for fundamental research to exploit available computational resources and more advanced analytics together. Recent progress has been based on ideas drawn from several disciplines e.g., computer science, engineering, mathematics, statistics, biology, medicine etc.
NSF's Information Technology Research (ITR) program, which stresses innovative and multidisciplinary approaches for significant problems, would be a good vehicle for funding basic research in pattern discovery. Some technical issues are: High dimensional data modeling, novel clustering techniques, combination of ensemble of classifiers, methods for scaling up supervised learning algorithms. Active and reinforced learning, learning and inference associated with graphical models including factor graphs, MRFs. Kernel machines, transductive inference techniques. Deformable templates; (time) sequential and 3D pattern recognition. Use of (stochastic) grammars. Fusion of multi-modal data from heterogeneous sensors in complex pattern recognition tasks; novel fusion paradigms. Current applications may include: Biometrics e.g., face, fingerprint, voice, iris, gait recognition. Tracking and surveillance including bio-surveillance possibly arising from epidemiological considerations e.g., spatio-temporal event detection. Recognition and discovery of information arising from fusion of data obtained via e.g., global sensing systems, biological, infrared and other detectors. Biomedical and financial forecasting applications may also be considered. In order to facilitate review and proper tracking of proposals, submission should be accompanied by the following designation of NSF program unit: Large ITR projects: CISE directorate. Medium ITR projects: CISE directorate, CCR division Small ITR projects: CISE directorate, CCR division, "Special Project" as a program unit. After submitting a proposal via FastLane, please let us know the proposal number by sending an email to: Sankar Basu Program Director, CISE/CCR Division National Science Foundation 4201 Wilson Blvd., Room 1145 Arlington, VA 22230 Phone: 703-292-8910 Email: [EMAIL PROTECTED] ITR DEADLINES: Large projects: Mandatory pre-proposals November 18, 2002. Medium projects: February 12, 2003 Small projects: December 12, 2002. ITR 2003 Website: http://nsf.gov/pubs/2002/nsf02168/nsf02168.htm
