Investigations on the theory and practice of credal classification

This project is concerned with developing new and more general methods of pattern classification, which we call credal classifiers. These classifiers have the potential to deal with challenging problems in classification such as the robust treatment of scarce or incomplete samples, and more generally the modelling of states of ignorance about the data-generating process. The project will focus on analyzing the theoretical properties of credal classification and on the development of new credal classifiers, as well as on their application to real problems.