Graph Neural Networks for Node Classification on Contact Networks

Understanding how infectious diseases or information spread through a network is critical for modern public health and network science. Applying machine learning to these interconnected structures requires specialized techniques beyond traditional tabular data models. This text-only course guides you through the process of modeling network connections and using Graph Neural Networks (GNNs) to classify nodes, such as identifying high-risk individuals in a contact tracing network. You will gain a solid conceptual and practical foundation in graph machine learning. What you'll learn: - Understand foundational graph theory concepts, node attributes, and adjacency matrices. - Build synthetic contact tracing graphs to simulate real-world transmission networks. - Explore modern Graph Neural Network architectures, including Graph Convolutional Networks (GCNs). - Prepare graph data and node features for machine learning pipelines. - Train GNN models to perform node classification tasks on imbalanced datasets. - Evaluate model performance using specialized graph-based metrics. You will start with the basic terminology of graphs and node attributes before moving step-by-step through synthetic data generation, network construction, and GNN model training. The written explanations and code walkthroughs ensure you understand both the theory and the implementation. This course is designed for beginners in machine learning and data science who want to expand their skills into graph-based data analysis, with no prior experience with graph neural networks required. Start reading today to unlock the power of graph machine learning for network analysis.