Hypothesis Testing on Random Networks

Speaker: 
Sebastien Bubeck, Princeton University – ORFE
Date: 
Feb 16 2015 - 4:30pm
Event type: 
PACM Colloquium
Room: 
Fine 214
Abstract: 

I will discuss some new hypothesis testing problems on random graph models. A popular example is the problem of detecting community structure in a graph. Here we will consider more exotic situations, such as testing one of the basic assumption in social network analysis: whether the graph has "geometric structure". We will also talk about dynamical models of random graphs (such as preferential attachment), and how to test different hypotheses on the "history" of these graphs.