Normalized Random Measures and Extensions

Abstract:
The normalized random measure is a large class of random probability
measures that has appealing properties and wide applications in
Bayesian nonparametrics. In this talk I will review the Poisson
process, completely random measures, and how to construct normalized
random measure from them. I will then talk about its posterior
characterization, and how to construct dependent normalized random
measures from it, e.g., spatial normalized random measures and mixed
normalized random measures. I then will show the applications of the
dependent normalized random measures in topic models. Some other
potential applications will be discussed as well.

Bio:
Changyou Chen is currently a PhD candidate in College of Engineering
and Computer Science, the Australian National University. He is also
affiliated with the Statistical Machine Learning (SML) group at
National ICT Australia (NICTA). He has a broad research interest in
machine learning, but is now focusing on Bayesian nonparametrics.