you can phisically find me (most of the times) at
School of Informatics, University of Edinburgh
3.18A Informatics Forum
10 Crichton Street Newington
Edinburgh EH8 9AB, UK



"quare id inferam fortasse requiris"

Can a single probabilistic model deliver robust and tractable inference
in presence of uncertainty over inputs (e.g., missing values, noise,...)
and detect anomalous values while dealing with heterogeneous
(mixed continuous and discrete, different likelihood models and statistical data types)?
Can we use it for exploratory data analysis?
Which kind of (probabilistic) patterns will it discover?