Predictive Models for Dynamic Causal Relationships in Network Structures

EasyChair Preprint 15030

Abstract

Network structures, commonly seen in social networks, biological systems, and economic markets, exhibit complex interdependencies that evolve over time. Understanding and predicting the dynamic causal relationships within these networks is crucial for various fields such as epidemiology, finance, and communication systems. Predictive models, particularly those leveraging machine learning and statistical inference techniques, have emerged as powerful tools to analyze such dynamic systems.

This paper focuses on developing and evaluating predictive models tailored to capture dynamic causal relationships in evolving network structures. Key challenges addressed include identifying latent variables, accounting for time-varying dependencies, and incorporating noise and uncertainty in large-scale networks. Techniques such as Granger causality, dynamic Bayesian networks (DBNs), vector autoregressive (VAR) models, and more advanced machine learning approaches like recurrent neural networks (RNNs) and graph neural networks (GNNs) are explored.

Keyphrases: Bayesian networks, Granger, Graph Neural Networks, causal inference, causality, dynamic networks, machine learning, predictive modeling

@booklet{EasyChair:15030,
  author    = {Wayzman Kolawole},
  title     = {Predictive Models for Dynamic Causal Relationships in Network Structures},
  howpublished = {EasyChair Preprint 15030},
  year      = {EasyChair, 2024}}