This project is about learning graphical models from data. Besides the general problem of learning a model, the group focuses on learning models for two particular subtasks within this domain, namely data clustering and data classification. In data clustering we are given data in the form of a set of instances with an (unknown) underlying group-structure, and the task is then to find the best description of this group-structure according to a certain criterion. Among the different definitions, interpretations, and expectations that the term data clustering gives rise to, we focus on a probabilistic or model-based approach to data clustering rather than on a partitional approach. In the related field of data classification, we have information about the group structure for a set of instances, and the task is then to learn a model for predicting the group membership for future instances.