We are pleased to announce that our visitor, David Tena Cucala, will be delivering a talk titled Expressive Power of Monotonic Graph Neural Networks.

When and where
– Date: October 7th
– Time: 12:00
– Location: NU-3B19

David Tena Cucala’s visit is funded by NI, and his talk promises to shed light on the complexities and expressive capabilities of Graph Neural Networks (GNNs).

Abstract
Graph Neural Networks (GNNs) are often used to learn transformations of graph data. While effective in practice, the expressive power of GNN-based approaches is often difficult to understand. In this talk, I will summarise recent work into monotonic GNNs, a sub-family of GNNs that are subject to restrictions ensuring that no prediction made by them is lost when extending their input. I will show that these GNNs correspond to a class of programs expressed in Datalog, a well-known rule-based formalism, and I will illustrate some applications of this relationship in the areas of verification and explainability.