School on Graph Neural Network and Explainable AI @ Liverpool

Published on

On September 20-22, a School on Graph Neural Network and Explainable AI has been organised at Liverpool by Prof. Monica D’Onofrio, Dr Cristiano Sebastiani and Dr Joe Carmignani, funded by the ERC CHIST-ERA project Multi-disciplinary Use Cases for Convergent new Approaches to AI explainability MUCCA. Around thirty scientists and students from particle physics, cosmology, accelerator and computer science gathered together to learn how to design graph neural networks and then move to explainability methods, from saliency maps to data attributions. Concrete examples have been discussed spanning from fundamental particle physics to medical applications and neuroscience.  Participants were invited to present briefly a project they are working on for which AI methods could be/are applied and, through an active learning approach, and could discuss with experts most suitable AI and XAI methods for their science case. Professor Pietro Lio (Prof in Computational Biology in the AI division, member of the Cambridge centre of AI and Medicine and The European Laboratory for Machine Learning) was invited as keynote speaker, and delivered a presentation of AI in medicine, focusing on how to build a digital patient twin using graph and hypergraph representation learning and considering physiological (cardiovascular), clinical (inflammation) and molecular variables (multi omics and genetics). A nice dinner complemented the school experience!