Bayesian nonparametrics for sparse dynamic networks

Conference item

In this paper we propose a Bayesian nonparametric approach to modelling sparse time-varying networks. A positive parameter is associated to each node of a network, which models the sociability of that node. Sociabilities are assumed to evolve over time, and are modelled via a dynamic point process model. The model is able to capture long term evolution of the sociabilities. Moreover, it yields sparse graphs, where the number of edges grows subquadratically with the number of nodes. The evolution of the sociabilities is described by a tractable time-varying generalised gamma process. We provide some theoretical insights into the model and apply it to three datasets: a simulated network, a network of hyperlinks between communities on Reddit, and a network of co-occurences of words in Reuters news articles after the September 11th attacks.

Authors: Naik, C; Caron, F; Rousseau, J; Teh, YW; Palla, K

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: Springer Nature
• Host title: Machine Learning and Knowledge Discovery in Databases. ECML PKDD 2022
• Pages: 191-206
• Series: Lecture Notes in Computer Science
• Series number: 13717
• Place of publication: Cham, Switzerland
• Publication date: 2023-03-17
• Acceptance date: 2022-06-18
• Event title: European Conference, ECML PKDD 2022
• Event series: ECML PKDD: Joint European Conference on Machine Learning and Knowledge Discovery in Databases
• Event location: Grenoble, France
• Event website: https://2022.ecmlpkdd.org/
• Event start date: 2022-09-19
• Event end date: 2022-09-23
• DOI: 10.1007/978-3-031-26419-1_12
• EISSN: 1611-3349
• ISSN: 0302-9743
• EISBN: 978-3-031-26419-1
• ISBN: 978-3-031-26418-4
• Language: English
• Keywords: stat.ML; point processes; Bayesian nonparametrics; sparsity; networks; random graphs; FFR; Poisson random measures