The National Science Foundation (NSF) awarded a $111,807 Project Grant under the Mathematical and Physical Sciences (CFDA 47.049) federal grant program to the University of Pittsburgh to develop robust algorithms for quickest threat detection in non-stationary, multi-stream data. The 3-year project aims to create provably robust algorithms that can quickly detect changes in the statistical properties of complex, time-varying data streams with unknown probability distributions. These algorithms will have applications in areas like public health monitoring, cyber-physical systems, and video surveillance. The project is divided into four technical thrusts to address challenges such as high-dimensionality, energy efficiency, and privacy concerns. The algorithms developed will be validated on public datasets and the code will be made openly available. The award provides research opportunities for students, with efforts to recruit from underrepresented groups.
Generated 1/28/25, 9:19 AM