|Talks|

Maximum-entropy temporal networks

Hybrid
Past Talk
Harrison Hartle
Aug 3, 2022
11:00 am
Aug 3, 2022
11:00 am
In-person
4 Thomas More St
London E1W 1YW, UK
The Roux Institute
Room
100 Fore Street
Portland, ME 04101
Network Science Institute
2nd floor
Network Science Institute
11th floor
177 Huntington Ave
Boston, MA 02115
Network Science Institute
2nd floor
Room
58 St Katharine's Way
London E1W 1LP, UK

Talk recording

This talk will be hybrid in-person and remote.

An important question for temporal network modeling is that of null model selection. We address this problem by applying the principle of maximum entropy to temporal networks. This allows the derivation of probability distributions on graph-sequences that are as unbiased as possible while satisfying a chosen set of constraints, be they exact or expectation-based. We investigate the many options for constraint-choices, and show how such choices impact the resulting dynamic networks. Constraints can be devised such that the model simultaneously yields (a) tunable network properties within each snapshot, and (b) tunable dynamic properties relating to graph-structure across multiple snapshots. Depending on the form of the latter constraint-types, the resulting temporal networks can be sequences of independently sampled graphs, first-order Markov chains, higher-order Markov chains, or altogether non-Markovian. We construct such models in both discrete and continuous time, with time-homogeneous and time-inhomogeneous constraints, studying them analytically and in large-scale simulations. We show that only certain combinations of constraints are allowed, and explore the parameter spaces of valid constraint-value combinations. We derive maximum-entropy temporal versions of well-known static network models, study their properties, compare them to various well-known temporal network models, and apply this framework to data from several real-world temporal networks. Additionally, we highlight interesting questions and remaining challenges.

About the speaker
Share this page:
Aug 03, 2022