skip to content
 

Inference in the infinite relational model

Presented by: 
Mikkel Schmidt Technical University of Denmark
Date: 
Wednesday 27th July 2016 - 14:30 to 15:00
Venue: 
INI Seminar Room 1
Abstract: 
The infinite relational model and similar related non-parametric mixture models are very powerful for characterizing the structure in complex networks. But (approximate) inference can be challenging, especially for large networks, and both MCMC and variational inference is often hampered by local optima in the posterior distribution. These issues are discussed in the context of learning a high resolution parcellation of human brain structural connectivity network.
University of Cambridge Research Councils UK
    Clay Mathematics Institute London Mathematical Society NM Rothschild and Sons