CHALKS is a research project devoted to applying topological approaches to enhance the explainability of models in Machine Learning and Deep Learning. The main goals are the study of the evolution of the latent space through Artificial Neural Networks and the relationship between the topological complexity of data and the architecture. Those goals have implications on architecture choice, model explainability and interpretability and similarities and dissimilarities between trained models. This research project will be carried out in three different institutions, which are the Graduate School of Mathematical Sciences at the University of Tokyo and the Mathematical Institute at the University of Oxford during the outgoing phase, and the University of Sevilla in the return phase.

Call: HORIZON-MSCA-2023-PF-01

Duration: 36 months

Grant agreement ID 101153039

DOI: 10.3030/101153039