Inferring dynamics of discrete states in time varying networks
Abstract
A collective of mobile agents forms a network based on physical proximity relationships. When they are close, they can communicate to exchange information, which may be useful for the benefits of the individual agents and the whole collective. Owing to the mobility, such proximity networks vary over time. There is an increasing amount of data on time varying networks derived from mobility or otherwise, stimulating invention of new algorithms for and mathematical understanding of time varying networks. The objective of the proposed project is to develop analysis pipelines to summarize such complex dynamics of networks as dynamics between a small number of discrete states. Given a time varying network, the proposed algorithms will infer the number of states and which state the system belongs to at any time point in principled manners. An inferred state may correspond to a normal state for the collective. Another inferred state may correspond to a paradigm shift (such as a new formation of vehicles) for the same collective. We will use Markov modeling approaches and multidimensional scaling to build our algorithms. Based on the proposed methods, we will also develop methodology to detect early warning signals in temporal network data. The aim is to detect a signature of a considerable change in network dynamics some time before the change occurs. We will deploy the proposed methods to empirical social temporal network data as well as synthesized temporal network data generated by mathematical models to verify their effectiveness. To create knowledge from the new methods, we will also actively interpret the obtained states and transitory dynamics between different states.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 14, 2022
- Source ID
- FA95501917024
Entities
People
- Alan R Champneys
Organizations
- Air Force Office of Scientific Research
- United States Air Force
- University of Bristol