Explainable Graph Neural Networks for Fraud Detection: Integrating XAI into Graph-Based Machine Learning Models for Financial Security - Softcover

Inhaltsangabe

This book presents a comprehensive study on the integration of Graph Neural Networks (GNNs) with Explainable Artificial Intelligence (XAI) methods for financial fraud detection. It evaluates multiple GNN architectures such as GCN, GAT, GIN, GraphSAGE, HinSAGE, and FraudGNN, alongside traditional machine learning models like Neural Networks and Random Forest. Explanation methods including GNNExplainer, GraphMask, SHAP, and LIME are applied to provide transparency, interpretability, and trust in fraud detection tasks. The work offers systematic comparisons in terms of performance, fidelity, runtime, and interpretability, supported by visual case studies. It highlights how combining graph-based reasoning with explainability techniques can improve fraud detection systems and meet emerging requirements of trustworthy and responsible AI.

Die Inhaltsangabe kann sich auf eine andere Ausgabe dieses Titels beziehen.