Back
2023INF008
Doctoral thesis

Latent drivers for dynamic networks

  • 2023

PhD: Università della Svizzera italiana

Dynamic networks
Latent space
Graph embedding
Dynamic node encoding
Time-stamp interactions
English Over the past few decades, network analysis has gained popularity in various fields, and understanding the dynamics of networks has become crucial. This thesis explores the dynamics of networks through a statistical approach, focusing on latent drivers that underlie network evolution. The thesis builds upon various key projects, each of which explores different aspects of network dynamics. The first project proposes a statistical testing procedure to determine whether the degree distribution of a given network follows a preferential attachment process, i.e., a power-law marginal distribution. The second project focuses on dynamic networks where the relational events constitute time-stamped edges and proposes a dynamic latent space relational event model, leveraging a Kalman filter EM algorithm. The third project extends it and addresses the challenge of dealing with huge relational event networks using machine learning optimization tools. The three projects investigate the complex phenomenon of network growth and transformation, shedding light on the role of latent drivers that shape the structure of observed networks. By studying the underlying drivers, analysts can better understand how networks impact various domains.
Collections
Language
  • English
Classification
Computer science and technology
License
License undefined
Open access status
green
Identifiers
Persistent URL
https://n2t.net/ark:/12658/srd1326508
Statistics
Document views: 43 File downloads:
  • 2023INF008.pdf: 73