The Bridge Between Graph Theory and Deep Learning via Graph Neural Networks

The Bridge Between Graph Theory and Deep Learning via Graph Neural Networks

Mathematics

Arjun Ashok

2025

Prof. Chenlei Leng

This project is about the use of Graph Neural Networks (GNNs) to understand the link between graph theory and deep learning. One of the main goals is to understand how GNNs can depict useful patterns in graph-structured data. I analysed the Elliptic Bitcoin transaction dataset which is used for fraud detection. By implementing a node classification model, the project shows how combining information from neighbouring nodes improves prediction accuracy compared to traditional machine learning models. The research illustrates the potential of GNNs in financial forensics and provides insight on model assessment with various metrics such as precision, recall, and F1-score.

Graph Neural Networks, Deep Learning, Graph Theory, Machine Learning, Artificial Intelligence, Node Classification, Fraud Detection, Cryptocurrency, Bitcoin, Elliptic Dataset, Data Science, Warwick URSS, Statistics