Bayesian Nonparametrics is characterised by “big” parameter spaces and the construction of probability measures on them. For the specification of these (random) probability measures there are mainly two general procedures: through the specification of their law or through a direct construction. For this seminar I will present an approach based on the normalisation of completely random measures that provides a direct construction and that leads to intuitive posterior structures.
Details about the programme of seminars organised by SMSAS postgraduate research students for SMSAS postgraduate research students.