Explore the propagation of uncertainty in network summaries through this 42-minute lecture by Eric Kolaczyk from Boston University, presented at the Alan Turing Institute. Delve into complex networks, noise characterization, and estimation techniques, including the Skellam distribution and normal approximations. Examine dependent errors, general treatments, and practical applications in gene coexpression networks. Gain insights into statistical analysis of network-indexed data, with a focus on methodology development and interdisciplinary applications in bioinformatics, computer science, geography, neuroscience, and sociology.
On the Propagation of Uncertainty in Network Summaries - Eric Kolaczyk, Boston University
Alan Turing Institute via YouTube
Overview
Syllabus
Intro
Outline
Complex Networks
But What About the Noise?!
Propagation of Noise: Network Summaries
Focus of this Talk: Characterization & Estimation
A General Formulation of the Problem
Assumptions
Key Observation
The Skellam Distribution
What about a normal approximation?
Dependent Errors
A General Treatment
Illustration: A Result for Dependent Noise
Reformulating the Problem
Is Estimation Possible in this Setting?
What If There Are Replicates?
Simulation Results
Application to a Gene Coexpression Network
Application: Results
Extensions
Some Closing Thoughts
Taught by
Alan Turing Institute
