Topics in Nonparametric Inference for unlabeled network models (Neyman Seminar)

Topics in Nonparametric Inference for unlabeled network models (Neyman Seminar)

Neyman Seminar
Nov 14, 2012, 04:00 PM - 05:00 PM | 1011 Evans Hall | Happening As Scheduled
Peter Bickel, Department of Statistics, University of California, Berkeley. (Speaker)
In B. and Chen PNAS(2009), we introduced a “nonparametric”model for unlabeled graphs with L=average degrees ranging from >>log(n) to O(n) and studied it in relation to “block models”. We will give an overview of our work in this regime including new results on maximum and variational likelihood for block models as well as methods generally appropriate for L->oo including methods presented in...