Graph Neural Networks (GNNs) & Relational Machine Learning (Relational ML)

Abstract

Graph Neural Networks (GNNs) have emerged as a powerful paradigm for processing non-Euclidean data structures, enabling machine learning models to exploit relational information in complex networks. Alongside, relational machine learning (Relational ML) has focused on learning predictive models over structured data with interdependencies between entities. This paper presents a comprehensive review of the foundational concepts, architectures, and recent advancements in GNNs and relational ML. We discuss methodologies such as Graph Convolutional Networks (GCNs), Graph Attention Networks (GATs), and Message Passing Neural Networks (MPNNs), as well as applications in social networks, recommender systems, and bioinformatics. Challenges, limitations, and future directions are also examined, emphasizing scalability, interpretability, and integration with symbolic reasoning. This review aims to provide researchers and practitioners with a consolidated understanding of graph-based learning frameworks.

KEYWORDS: Graph Neural Networks, Relational Machine Learning, Graph Convolutional Networks, Message Passing, Graph Attention Networks, Knowledge Graphs