Pietro Bongini
University of Siena - Dipartimento di Ingegneria dell'Informazione e Scienze Matematiche Niccolò Pancino
University of Siena - Dipartimento di Ingegneria dell'Informazione e Scienze Matematiche
Course Type
Type B
Calendar
7-11/9/2026
Room
Program
Graph Neural Networks are emerging as powerful deep learning models that can process almost any kind of graph-structured data. Their versatility guarantees a high learning potential on any kind of task, from classification and regression, to representation learning and graph generation. Since graph-structured data are ubiquitous in research as well as in the industrial domain, processing graphs more efficiently, extracting more useful information from them, and even generating graphs with desired characteristics are objectives of major importance to advance human knowledge. This course proposes an overview of the main GNN models, grouped into two broad families (Convolutional and Recurrent GNNs), with a particular focus on their applications and on the software tools to implement them.
