Probabilistic Graphical Models. Fundamental
representations and methods for inference and learning in large scale
domains, with an emphasis on high-level elements such as structure
learning, the discovery of hidden variables and classes, transfer of knowledge between
related classes/tasks. I have recently taken a particular interest to
nonlinear high-dimensional representation of continuous or hybrid distributions.
Real-life Applications. Applying fundamental techniques to
challenging domains such as computational biology and machine vision.
Recently, my I have started focusing on the development of principled
techniques based on probabilistic knowledge for diagnosis in the field
of medical informatics.