Explainable AI: Theoretical Models and Applications in Multi-attribute Heterogeneous Data

It is argued that explainability can be achieved most effectively through the use of appropriate data representations, such as heterogeneous multi-attribute representations, which preserve the original nature of the results.

In this context, this project focuses on two fundamental aspects. First, we undertake theoretical research to define, analyse and construct fusion functions that select representative data in problems with real and heterogeneous multi-attribute data. Second, we seek to apply this research to general machine learning or deep learning problems.

Project funded by: