Graphs have the ability to represent interactions of large and complex systems. While the study of static networks is wide and established, less attention has been paid to dynamic networks, despite most networks being dynamic in nature. In this work, we address the problem of dynamic network embedding, that is, representing the nodes of a dynamic network as evolving vectors within a low-dimensional space. This thesis details the creation of a new suite of tools to embed dynamic networks, interpret the structure encoded in dynamic network embeddings, produce predictions on dynamic networks with quantifiable uncertainty, and bootstrap networks to quantify the uncertainty in their representation.
| Date of Award | 18 Mar 2025 |
|---|
| Original language | English |
|---|
| Awarding Institution | |
|---|
| Sponsors | LV= General Insurance |
|---|
| Supervisor | Daniel John Lawson (Supervisor) & Patrick Rubin-Delanchy (Supervisor) |
|---|
- Graphs
- Machine learning
- Geometric Deep Learning
- Hypothesis Testing
- Bootstrap
- Conformal Inference
- Embeddings
- Graph Embedding
- Dynamic Graphs
Beyond Spectral Unfoldings for Dynamic Network Embeddings
Davis, E. J. (Author). 18 Mar 2025
Student thesis: Doctoral Thesis › Doctor of Philosophy (PhD)