|
Abstract:
|
With network data becoming ubiquitous in many applications, many models and algorithms for network analysis have been proposed, yet methods for providing uncertainty estimates are much less common. Bootstrap and other resampling procedures have been an effective general tool for estimating uncertainty from i.i.d. samples, but resampling network data is substantially more complicated, since we typically only observe one network. We consider several fully nonparametric network resampling procedures based on vertex and edge sampling, and some alternatives based on low-rank and latent variable models. We find that no one procedure is best for all tasks, and demonstrate the pros, cons, and various trade-offs of different bootstrapping strategies.
Based on joint work with Qianhua Shan and Keith Levin.
|
