TITLE:
Estimating Network Parameters
ABSTRACT:
Estimating the parameters of a large network, which cannot be accessed in its
entirety, is a basic statistical question. In this talk we address two
questions in this context. First is that of estimating the average degree of
a network; we show a simple non-uniform sampling algorithm that can do this
using a small number of samples, which is better than the most obvious
estimator for this problem. Second is the question of estimating the fraction
of nodes with a certain property; here, we show that an easy estimator can be
improved upon by using information about the neighborhood of a node.