Contagion Origins: Maximum-Likelihood Estimation in Spreading Models
Core Concepts
Probabilistic approach for rumor source detection using maximum likelihood estimation in spreading models.
Abstract
The article discusses the importance of identifying the source of epidemic-like spread in networks, focusing on rumor source detection. It introduces a probabilistic approach utilizing maximum likelihood algorithms and starlike tree approximations for general graphs. The study evaluates the effectiveness of the proposed algorithm in diverse network scenarios, advancing information dissemination strategies.
I. Introduction
Epidemic-like spreading is crucial in network science.
Malicious information proliferation poses cybersecurity challenges.
COVID-19 infodemic triggered an epidemic of online misinformation.
WHO declared war against COVID-19 Infodemic.
II. System Model
Rumor-spreading model outlined with maximum likelihood estimator.
Rumor source estimator based on observed graph data and precise timings.
III. Rumor Source Detector for Tree Graphs
Evaluation of likelihood ratio between nodes in tree graphs.
Theoretical results for d-regular trees and likelihood ratios between adjacent nodes.
IV. Approximations and Asymptotics by Starlike Tree Graphs
Starlike tree graphs defined and approximation algorithm presented.
Analysis of asymptotic behavior of likelihood ratio in starlike trees.
V. Experiments
Numerical examples illustrate properties of the proposed estimator on starlike trees.
Comparison with existing methods like rumor centrality and distance centrality in Random Trees and ER random graphs.
VI. Further Discussions
Markovian models like hyper-Erlang distributions and phase-type distributions can enhance understanding of rumor dynamics.
Unraveling Contagion Origins
Stats
"Comprehensive evaluations confirm algorithmic effectiveness."
"Gamma functions offer potent tool for analyzing graph-theoretic features."
"Asymptotic behavior analyzed for likelihood ratio between nodes."
Quotes
"The challenge lies in tracing the source from a snapshot observation."
"Novel starlike tree approximation extends applicability to general graphs."
"Our formulation provides valuable insights into network resilience."