|
Activity Number:
|
174
- Dynamic Network Modeling
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 31, 2017 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistical Learning and Data Science
|
|
Abstract #324624
|
|
|
Title:
|
A Fast Scalable Procedure for Finding Change Points in Random Graphs
|
|
Author(s):
|
Mingyuan Gao* and George Michailidis
|
|
Companies:
|
University of Florida and University of Florida
|
|
Keywords:
|
change point ;
inference
|
|
Abstract:
|
We consider the problem of estimating a change point in a sequence of random graphs. We propose a computationally efficient two stage procedure, where at the first stage the algorithm finds a pilot estimator via a specially designed sampling scheme on the network edges, and at the second stage the algorithm obtains the final estimator based on the pilot estimator. It can be shown that such procedure can achieve the same asymptotic performance as the estimate based on all the data. Experimental results based on synthetic data show that this idea can effectively reduce computational burden in several different settings.
|
Authors who are presenting talks have a * after their name.
