Zapraszamy na seminarium prof. Jorge Mateu

Już w środę 8 listopada odbędzie się seminarium "Statistical models and deep learning methods for the analysis, prediction and monitoring of space-time point pattern data".

Zapraszamy o godz. 15:00 do s. A203.

Abstract: We present several statistical approaches to understand the underlying temporal and spatial dynamics of events evolving in space and time that can result in informed and timely public policies. We focus on analysing high-resolution data in form of spatio-temporal point patterns,  offering vital insights for the spatio-temporal evolution of events linking it with their spread in a region.
We develop a batery of models and approaches, ranging from non-stationary spatio-temporal point processes, mechanistic models giving particular data-driven forms to the spatio-temporal intensity function, stochastic covariance-based models,  cluster spatio-temporal models to identify unknown sources, and methods of spatial growth functions able to develop velocities of the spread of the events. The idea of barycenter in spatial point patterns is also delineated.
We additionally provide a mathematical framework for coupling neural network methods with the statistical analysis of point patterns with a focus on two problems. We first use deep convolutional neural networks and employ a Siamese framework to build a discriminant model for distinguishing structural differences between spatial point patterns. Then, we discuss an example of deep neural networks in the statistical analysis of highly multivariate spatial point patterns and provide a new strategy for building spatio-temporal point processes using variational autoencoder generative neural networks.