ORIE Professor Resnick leads team that garnered a major grant from the Department of Defense

Multivariate heavy tail statistics, identified as a key area of scientific inquiry, is the subject of a DoD grant to Sidney Resnick’s interdisciplinary team.

Two years ago the US Department of Defense (DoD) identified 21 areas of scientific inquiry and solicited research proposals in these areas as part of their Multidisciplinary University Research Initiative (MURI). 

One key area is multivariate heavy tail statistics, a subject of current research in ORIE; others in the broad program range from “quantized chemical reactions of ultracold molecules” to “predictive models of cultural and behavioral effects of societal stability.”  A team headed by ORIE’s Teng Hui Lee Professor Sidney Resnick and including ORIE Professor Gennady Samorodnitsky has now been awarded a multimillion dollar MURI grant for basic research on multivariate heavy tail statistics.  The funding will be used to support graduate students as well as faculty research.

These awards have been made by the DoD MURI program, which is designed to support research spanning multiple science and engineering disciplines and multiple academic institutions.   Resnick’s team is from seven universities, involving faculty of the University of Massachusetts, Columbia, American, Ohio State, the University of Illinois and the University of Minnesota in addition to Cornell.   His team members are in the fields of electrical and computer engineering, computer science, statistics, and mathematics in addition to operations research.

Heavy tailed phenomena are those for which the extreme events have much higher probability than would be predicted by the Gaussian (“bell curve”) model. For example a heavy tail distribution can be used to model the number of Twitter followers that an individual has.  When a heavy tailed phenomenon is characterized by more than one variable, such as the number of followers of individual members of a group, understanding how extreme values of the variables might be depend on one another can be critical.  

However traditional measures of dependence such as regression and correlation are inappropriate in such cases – “they just don’t work,” according to Resnick, “because the fitted probability distributions in question actually don’t have a variance or standard deviation, values on which the traditional measures rely.”   Analyzing social networks, detecting anomalous behavior of communications networks, and maintaining secrecy and anonymity on wireless networks are examples where better methods based on understanding dependence in multivariate heavy tail phenomena are needed. 

Cornell is also represented on another MURI team.   Tisch University Professor Jon Kleinberg (a member of the Field of Operations Research) and Goldwin Smith Professor of Economics Lawrence Blume are on a team awarded a MURI grant for basic research on “cultural norms and the dynamics of socio-political change,” in response to the topic on cultural and behavioral effects of societal stability.  Their team is headed by Professor Ali Jadbabaie of the University of Pennsylvania.    

Other Articles of Interest