Artificial intelligence applied to graphs: Graph Neural Networks

Graphs are a set of objects (called nodes) and the relationships between them (edges). Graphs are very popular for representing maps, molecules or the connections in a computer network and other applications. However, until recently there were no neural networks capable of working with graphs, which has limited the applicability of artificial intelligence in many industrial fields. Graph Neural Networks (GNN) were put forward recently to solve this problem and have revolutionised many fields of knowledge.

In this masterclass we will see what GNNs are and their most popular applications. Plus through a PRACTICAL DEMO, we will implement a GNN using the IGNNITION framework (https://ignnition.org). This framework does not require programming knowledge in Tensorflow or Pytorch.

Addressed to:

People interested in graphs and AI.
A basic knowledge of mathematics and AI is required to follow the talk.

Programa

– What Graph Neural Networks are
– Applications of Graph Neural Networks
– Practice with a Graph Neural Network using the IGNNITION framework (https://ignnition.org).

Masterclass taught by IDEAI-UPC

Play video Download Presentation

Taught by:

Albert Cabellos, professor at the Polytechnic University of Catalonia, member of IDEAI.
Albert Cabellos is a professor in the Department of Computer Architecture at the Polytechnic University of Catalonia, ICREA and a member of IDEAI. Albert has been a visiting professor at MIT (Boston), UC Berkeley (San Francisco), KTH (Stockholm) and NII (Japan) as well as a visiting scientist at Cisco and Agilent Technologies.
Albert researches how to apply AI to telecommunications infrastructures.
Miquel Ferriol, PhD student at the Polytechnic University of Catalonia, member of IDEAI. Miquel Ferriol-Galmés received his B.Sc. degree in Computer Science and his M.Sc. degree in Data Science respectively in 2018 and 2020 from the Polytechnic University of Catalonia (UPC). He is currently pursuing a PhD at the Barcelona Neural Networking Center (BNN-UPC). His main research interests are in the application of Graph Neural Networks to network modelling and optimisation, as well as in the application of blockchain to networks.