Italiano (Italian) English (Inglese)
Friday, 22 January 2021


End Date: 31/03/2020

Talk Announcement

Set-valued quantification of probabilistic graphical models with applications to machine learning and decision-support systems.
Thu 26/3/2020 ore 14.30, Seminar Room, Computer Science (C192).
Speaker: Alessandro Antonucci, Senior Lecturer-Researcher at the 'Dalle Molle' Institute for Artificial Intelligence (IDSIA)
Abstract: Credal networks are a generalisation of Bayesian networks, where local distributions are replaced by convex sets of them. This gives higher expressiveness to the models without preventing the possibility of a compact specification of a joint generative model, as well as the possibility of performing reasoning by inference algorithms. We present such a class of model and discuss the existing inference tools available for them. Applications to machine learning are presented together with a number of knoledge-based decision-support systems based on these models.