Description: Description: Worcester Polytechnic Institute (WPI)

Description: Description:


Space-time signal processing for distributed pattern detection in sensor networks

 Dr. Randy Paffenroth
Numerica Corporation



In this talk we will present theory and algorithms for detecting weak, distributed patterns in large scale network data.  The patterns we consider are anomalous temporal correlations between signals recorded at sensor nodes across a network.  We use robust matrix completion and second order analysis to detect distributed patterns that are not discernible at the level of individual sensors.

When viewed independently, the data at each sensor cannot provide a definitive determination of the underlying pattern, but when fused with other sensors the relevant patterns emerge.  The approach has been applied to several domains, including financial analytics, crime analytics, and chemical detection, where correlated phenomena are of interest.  Our focus in this talk will be on detecting weak patterns in computer networks where the nodes (terminals, routers, servers, etc.) are sensors that provide measurements (of packet rates, user activity, CPU usage, etc.). Further, we demonstrate how these methods can be efficiently implemented and applied to large distributed databases.



Dr. Paffenroth graduated from Boston University with degrees in both mathematics and computer science.  After defending his thesis, he was awarded his Ph.D. in applied mathematics from the University of Maryland in 1999.  After attaining his Ph.D., Dr. Paffenroth spent seven years as a Staff Scientist in Applied and Computational Mathematics at the California Institute of Technology.  In parallel with his position at the California Institute of Technology he was also a Scientist at MathSys Inc.  In 2006 he joined Numerica and has since held the position of Computational Scientist and, most recently, Program Director. As Program Director, Dr. Paffenroth leads a team of scientists (B.A., M.A., and Ph.D.) as well as performing mathematical research, software engineering, and proposal development. Before joining the Numerica team, Dr. Paffenroth developed theory and software for high performance computing applications including high-order methods for boundary integral formulations of PDEs and bifurcation analysis in ODEs.  His current technical interests include large scale machine learning, signal processing, compressed sensing, and the interaction between mathematics, computer science and software engineering.  Dr. Paffenroth's work has been funded by a number of agencies including the US Air Force, the US Army, the US Navy, the Defense Threat Reduction Agency, NASA, the Department of Homeland Security, the Office of the Secretary of Defense, and various industrial partners.  He also holds various honorary positions including a faculty affiliate position in Electrical and Computer Engineering at Colorado State University, a visiting Scholar position in Electrical, Computer, and Energy Engineering at University of Colorado, Boulder, and he sits on the ECE Industrial Advisory Board at Colorado State University.

Host:  Professor Craig Wills

Refreshments will be served.