A Spectral Framework for Graph Sampling

Abstract

Using techniques from Markov chains, characterize the effective sample size of a Markov chain sample and relate this to the spectral characteristics of the graph. Use these results to estimate standard errors (for confidence intervals and hypothesis testing). Theoretically validate the consistency of these standard error estimates. Sub-aim1.2: Develop globally-adaptive Markov chains, prove that they are globally-adaptive, and extend the results from the previous sub-aim to these novel chains.Sub-aim1.3:Extend the results from the previous two sub-aims to more general tasks of statistical inference on node/edge contextualizing measures (e.g. two sample tests, linear regression, principal components analysis) and their relationships to network topology (e.g. degree and clustering coefficient).

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 18, 2021
Accession Number
AD1203439

Entities

People

  • Karl Rohe

Organizations

  • University of Wisconsin–Madison

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Algorithms
  • Computational Science
  • Data Science
  • Factor Analysis
  • Governments
  • Information Processing
  • Information Science
  • Machine Learning
  • Markov Chains
  • Markov Models
  • Markov Processes
  • Media
  • Probability
  • Random Walk
  • Simulations
  • Social Media
  • Social Networking Services
  • Social Networks
  • Statistical Analysis
  • Statistical Inference
  • Statistics
  • Stochastic Processes
  • Surveys

Readers

  • Neural Network Machine Learning.
  • Statistical inference.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Machine Learning Algorithms