Markov process models for time-varying networks: basic theory and practical limitations

Harry Crane - Rutgers University
Friday, March 3, 2017 -
3:30pm to 4:30pm
hcrane@stat.rutgers.edu
Old Chemistry 116

Abstract: 

I show some recent developments for modeling networks and related structures (such as multilayer networks) whose connectivity varies over time, as is common for networks representing social relationships and other kinds of temporally varying interactions.  I focus much of the discussion on how basic model assumptions, such as exchangeability and the Markov property, combine to restrict the class of models in a way that may be unappealing for many applications.  I then propose another class of models, Combinatorial Levy Processes (CLPs), which no longer require exchangeability but instead assume a certain stationarity in time (akin to real-valued Levy processes).  CLPs seem to have severe practical limitations of their own, raising the question: What (if any) of this theory be salvaged for practical purposes?



Series: 
Statistical Science Seminar Series

Description: 

Seminars generally take place in 116 Old Chem Building on Fridays from 3:30 - 4:30 pm. However, please check individual abstracts to confirm time and location. Refreshments will be served after the seminars in Old Chemistry 211. Metered Parking is available at various locations on campus. If you have never visited us before, please see our driving directions and map. The easiest and most convenient parking areas are located at the Bryan Center parking garage near Duke Statistics (recommended) or at the Sarah P. Duke Gardens. Please email or call Karen Whitesell for additional information: karen.whitesell@duke.edu or phone 919-684-8029. Sorry, but we do not have reprints available. Please feel free to contact the authors by email for follow-up information, articles, etc. Reception following seminar in 211 Old Chemistry