Mission Statement:

The focus of our knowledge discovery efforts is to develop actionable predictive insights from massive and disparate data, which may in turn may be generated from dynamic, temporally evolving and geographically distributed processes. Our fresh approach stem from two considerations. First, our research is geared towards high-priority application solutions for emerging national and societal priorities, which imply new challenges and lead to innovative solutions. For example, a key thrust area is the mitigation of natural or anthropogenic hazards, including but not limited to hurricanes and terrorist attacks, based on data from diverse sensors and computational models. Second, our approach to solutions is multidisciplinary, spanning traditional methodological disciplines like computational statistics, computer science and data mining, nonlinear dynamics, signal processing and econometrics, as well as high-performance computational modeling and simulation. Thus, when precursors or signatures of rare or extreme events and abrupt change need to be detected or modeled through innovative use of massive and disparate data, whether for weather extremes or for security alerts, non-traditional and completely new data scientific methodologies may need to be utilized and developed. Our knowledge discovery solutions are ultimately oriented towards the development of decision aids, including coordinated offline discovery and real-time analysis for faster and more reliable tactical decisions, as well as longer-term predictive approaches which can facilitate the effective design of strategic policy.

"Knowledge Discovery", Presentation by Brian Worley, CSE Division Director.

Recent News:

The CSE Division is co-sponsoring the 2nd International Workshop on Knowledge Discovery from Sensor Data, to be held on August 24th in Las Vegas in conjunction with the 14th ACM Conference on Knowledge Discovery and Data Mining. Additional information is available from the links below:
2nd International Workshop on Knowledge Discovery from Sensor Data (Sensor-KDD 2008)

High-level summaries of our work on knowledge discovery for security-related hazards and natural geophysical hazards have been submitted as book chapters in  "Learning from Data Stream: Processing Techniques in Sensor Networks". Book Chapter on Security-Related Hazards and Book Chapter on Natural Geophysical Hazards (coming soon...)

Our paper on technological hazards has been submitted to ACM Transaction on Sensor Networks: Huang, C., Hsing, T., Cressie, N., Ganguly, A.R., Protopopescu, V.A., and N.S. Rao (2006): Statistical Analysis of Plume Model Identification Based on Sensor Network Measurements.