Skip to content

References: Graph Embeddings, Clustering, and Graph Neural Networks

  1. Graph neural network - Wikipedia - Comprehensive overview of GNN architectures, message-passing layers, and graph convolutional networks, directly grounding this chapter's formalization of Chapter 5's manual neighbor-averaging into trainable, learned-weight embeddings.

  2. Node2vec - Wikipedia - Explains the biased second-order random-walk sampling and skip-gram training behind Node2Vec, supporting the chapter's walk-based alternative to message passing for generating node embeddings.

  3. Louvain method - Wikipedia - Describes the iterative modularity-optimization procedure and its two-phase local-move-then-aggregate structure, matching the chapter's worked trace of Louvain clustering a small cardiac/diabetes patient network.

  4. Graph Representation Learning - William L. Hamilton - Morgan & Claypool (Synthesis Lectures on AI and ML) - Hamilton is credited with the unifying encoder-decoder framework that places shallow random-walk embeddings (DeepWalk, Node2Vec) and message-passing GNNs side by side as two solutions to one problem, mirroring this chapter's "two philosophies, one goal" framing.

  5. Network Science - Albert-László Barabási (with Márton Pósfai) - Cambridge University Press - Barabási is widely credited for the field's most accessible diagrams and worked derivations of assortativity, network motifs, and community/modularity structure, freely available online and adopted as the standard visual reference for the structural metrics this chapter introduces.

  6. Louvain - Neo4j Graph Data Science Documentation - Official reference for running the Louvain modularity-optimization algorithm on a property graph, including configuration parameters and worked examples, directly usable for the healthcare cohort-clustering scenarios this chapter describes.

  7. Node2Vec - Neo4j Graph Data Science Documentation - Documents the random-walk parameters (return factor, in-out factor) and training process behind Node2Vec embeddings, extending the chapter's Maria Chen walk-trace example into a runnable graph-database algorithm.

  8. A Gentle Introduction to Graph Neural Networks - Distill.pub - Interactive, visually-driven explanation of how GNN message passing gathers, aggregates, and updates node embeddings, reinforcing the chapter's step-by-step derivation of the learned-weight GCN update rule.

  9. CS224W: Machine Learning with Graphs - Stanford University (Jure Leskovec) - Course site covering node embeddings, random walks, and graph neural network architectures at the depth this chapter introduces, useful for students wanting a deeper follow-up treatment of node classification and GCNs.

  10. Number of Triangles in an Undirected Graph - GeeksforGeeks - Walks through adjacency-matrix and bitset algorithms for triangle counting with complexity analysis, supporting the chapter's triangle-count metric used as the raw input to clustering-coefficient and cohort-density calculations.