PC Seminar: Stein’s method for exponential random graph models and assessing goodness of fit
- Date: 3 November 2022, 10:15–23:59
- Location: Ångström Laboratory, Seminar room in house 6 and Zoom
- Type: Seminar
- Lecturer: Gesine Reinert, Oxford University
- Organiser: Matematiska institutionen
- Contact person: Tiffany Lo
Welcome to this seminar held by Gesine Reinert, Oxford University with the title "Stein’s method for exponential random graph models and assessing goodness of fit ".
Abstract: Exponential random graph models are a key tool in network analysis but due to an intractable normalising constant are difficult to manipulate. In this talk we shall use Stein's method to approximate these models by Bernoulli random graphs in ``high temperature" regimes.
For assessing the goodness of fit of a model, often independent replicas are assumed. When the data are given in the form of a network, usually there is only one network available. If the data are hypothesised to come from an exponential random graph model, the likelihood cannot be calculated explicitly. Using a Stein operator for these models we introduce a kernelized goodness of fit test and illustrate its performance.
Finally, we extend the ideas of this goodness of fit test to provide an approximate goodness of fit test for potentially black-box graph generators.
This talk is based on joint work with Nathan Ross and with Wenkai Xu.
Participate on site or on Zoom (meeting ID: 5468070770, Password: 141592)
This is a seminar in our seminar series on Probability and Combinatorics (PC).