Agenda

Speaker:
Antonio Peruzzi, Ca' Foscari University of Venice - Department of Economics

Link Zoom

Abstract:
A new dynamic latent space eigenmodel (LSM) is proposed for weighted temporal networks. The model accommodates integer-valued weights, excess of zeros, time-varying node positions (features), and time-varying network sparsity. The latent positions evolve according to a vector autoregressive process that accounts for possible lagged and contemporaneous dependence across nodes and features, which is neglected in the LSM literature. A Bayesian approach is used to address two of the primary sources of inference intractability in dynamic LSMs: latent feature estimation and the choice of latent space dimension. Regarding the first task, we employ an efficient auxiliary-mixture sampler that performs data augmentation and supports conditionally conjugate prior distributions. A point-process representation of the network weights and the finite-dimensional distribution of the latent processes are used to derive a multi-move sampler in which each feature trajectory is drawn in a single block, without recursions. This sampling strategy is new to the network literature and can significantly reduce computational time while improving the mixing of the chain. To avoid trans-dimensional samplers, a Laplace approximation to a partial marginal likelihood is used to design a partially collapsed Gibbs. Overall, our procedure is general, as it can be easily adapted to static and dynamic settings, as well as to other discrete or continuous network weight distributions. We assess the accuracy of the proposed methodology through extensive simulation studies and illustrate its relevance with applications to brain connectivity and UN roll-call voting networks

Bio sketch:
Antonio Peruzzi is a Postdoctoral Researcher at Ca’ Foscari University of Venice, working at the intersection of Bayesian statistics, Econometrics, and Complex Networks. Previously, he worked as a Senior Scientist at Alpen-Adria Universität Klagenfurt. He holds a Ph.D. in Economics at Ca' Foscari University of Venice, where he worked under the supervision of Prof. Roberto Casarin. His current research focuses on Latent Space network models, with meaningful applications spanning diverse socio-economic domains, including Media Economics and Quantitative Finance.