Skip to content

About This Book

Welcome from Sage!

Sage waving welcome Hi, I'm Sage — a graph node who learns by aggregating wisdom from my neighbors, just like the GNNs we study together. This textbook started as a question: what would it look like if a student built the GNN course they wished existed? Anvith spent months building exactly that. Whether you're starting from zero or deepening existing knowledge, you're in the right place. Let's aggregate some knowledge!

Why This Intelligent Textbook

Graph-structured data is everywhere — social networks, drug interaction databases, knowledge bases, supply chains, financial transaction networks, and the molecular graphs underlying modern drug discovery. Graph Neural Networks have emerged as the definitive tool for reasoning over this class of data, yet they remain poorly covered in most ML curricula: either treated as a brief aside in a general deep learning course, or fragmented across research papers with no unified pedagogical thread.

Graph ML is already operating at planetary scale:

  • Pinterest's PinSage, a GNN-based recommender, generates recommendations from a graph of 3 billion nodes and 18 billion edges — one of the largest deployed GNN systems in production, responsible for driving 40% of home-feed impressions.1
  • AlphaFold 2, which solved the 50-year protein-folding problem and won the 2024 Nobel Prize in Chemistry, uses an attention mechanism over residue-pair graphs to achieve atomic accuracy — demonstrating that graph-structured reasoning operates at the frontier of scientific discovery.2
  • The global knowledge graph / graph analytics market was valued at $2.4 billion in 2023 and is projected to grow at a CAGR of 24.3% through 2030, driven by adoption in healthcare, finance, and enterprise AI.3
  • The number of papers referencing "graph neural network" on arXiv grew from under 100 in 2017 to over 7,500 in 2023 — a 75× increase in six years — making graph ML one of the fastest-growing subfields in machine learning.4
  • Drug discovery pipelines at AstraZeneca, Pfizer, and over 50 biotech startups now use GNNs to predict molecular properties, synthesize novel compounds, and repurpose existing drugs — with GNN-assisted compounds entering clinical trials as of 2024.5

These numbers represent a field that has already left the lab and is reshaping industries. Yet most students finish a general ML course without having written a single line of GNN code or seen a message-passing derivation from first principles.

This textbook closes that gap. It is built on a learning graph of 300 interconnected concepts organized across 15 taxonomy categories, so prerequisites are always introduced before they are needed. Every chapter leads with a real-world motivating example, builds intuition before equations, provides full mathematical derivations, and includes complete runnable PyTorch Geometric code. 37 interactive MicroSims — p5.js simulations embedded directly in the browser — let you manipulate GNN internals hands-on. The entire textbook is open source and free — no paywalls, no access codes, no expensive editions.

Audience

This textbook is designed for:

  • Graduate students in CS, computational biology, data science, or any field with relational data
  • Advanced undergraduates who have completed an ML course and want to go deeper on graphs
  • Practitioners building recommendation engines, drug discovery pipelines, fraud detection systems, or knowledge bases
  • Researchers entering graph ML who want a unified, up-to-date reference

Prerequisites

Required

  • Programming: Python proficiency; NumPy/Pandas familiarity
  • Linear algebra: Matrix multiplication, eigenvalues/eigenvectors, SVD
  • Probability & statistics: Probability distributions, expectations, Bayes' theorem
  • ML fundamentals: Supervised/unsupervised learning, gradient descent, backpropagation

Helpful but taught in Chapter 0

  • PyTorch and automatic differentiation
  • Graph theory basics (introduced from scratch)
  • PyTorch Geometric (PyG)

What Makes This Textbook Different

  • Self-contained prerequisites — Chapter 0 covers linear algebra, probability, PyTorch, and PyTorch Geometric from scratch. No assumed background beyond introductory ML.
  • Intuition before formalism — every concept is explained in plain language before equations appear.
  • Full derivations — major results are derived, not just stated.
  • Interactive MicroSims — 37 p5.js simulations let you manipulate parameters and see GNN behavior in real time.
  • Bloom's-aligned exercises — 12 exercises per chapter spanning all cognitive levels, from recall to open-ended design.
  • Up-to-date benchmarks — results from 2024–2025, with links to live leaderboards.
  • Broad coverage — 27 chapters covering both classical graph algorithms and frontier topics like graph foundation models and LLM+GNN integration.

How to Use This Book

This textbook is designed for self-paced study. The book includes:

  • 27 Chapters spanning graph theory, classical algorithms, node embeddings, GCN/GAT/GIN, graph transformers, knowledge graphs, heterogeneous graphs, scalable GNNs, generative models, and LLM+GNN integration
  • 37 Interactive MicroSims embedded in chapters — browser-based p5.js simulations you can manipulate to explore message passing, attention weights, WL refinement, PageRank convergence, and more
  • 270 Quiz Questions across all chapters, distributed over all 6 Bloom's Taxonomy levels
  • Annotated References per chapter linking to Wikipedia and authoritative primary sources
  • Glossary with definitions for all 300 key concepts
  • FAQ with 82 common questions and answers
  • Learning Graph visualizing 300 concept dependencies so you can plan a non-linear path

Navigation tips:

  • Read Chapter 0 first if you need a refresher on linear algebra, PyTorch, or PyTorch Geometric
  • Read chapters in order — concepts are introduced in dependency order per the learning graph
  • Use the search bar (top right) to jump to a specific term
  • Try the MicroSims as you encounter them — they are the fastest way to build intuition for a new concept
  • Check the Learning Graph when you want to see how a concept fits into the larger picture
  • Each chapter's "Further Reading" section points to the key papers — follow those after mastering the chapter content

About the Author

Anvith Pothula is a student and independent developer with a deep interest in machine learning, graph-structured data, and open education. This textbook grew out of a desire to build the comprehensive, interactive GNN resource that didn't yet exist — one that matches the depth of graduate coursework while remaining fully accessible to motivated undergraduates and self-taught practitioners.

The project was built end-to-end using MkDocs Material, p5.js, PyTorch Geometric, and Claude AI — demonstrating how a single determined builder can produce graduate-level educational content at no cost to learners.

Other projects:

  • Machine Learning Textbook — a companion interactive textbook covering the ML fundamentals (supervised learning, neural networks, optimization, evaluation) that this book builds on

Connect:

How to Cite This Book

If you reference this textbook in academic work, curriculum proposals, lesson plans, or other publications, please use one of the following formats.

APA (7th edition)

Pothula, A. (2026). Graph Neural Networks: An Intelligent Interactive Textbook. https://AnvithPothula.github.io/graph-neural-networks-textbook/

Chicago (17th edition)

Pothula, Anvith. 2026. Graph Neural Networks: An Intelligent Interactive Textbook. https://AnvithPothula.github.io/graph-neural-networks-textbook/.

MLA (9th edition)

Pothula, Anvith. Graph Neural Networks: An Intelligent Interactive Textbook. 2026, AnvithPothula.github.io/graph-neural-networks-textbook/.

BibTeX

@book{pothula2026gnn,
  title       = {Graph Neural Networks: An Intelligent Interactive Textbook},
  author      = {Pothula, Anvith},
  year        = {2026},
  version     = {0.1},
  url         = {https://AnvithPothula.github.io/graph-neural-networks-textbook/},
  doi         = {},
  keywords    = {graph neural networks, message passing, GCN, GAT, GIN, GraphSAGE,
                 graph transformers, knowledge graphs, PyTorch Geometric},
  license     = {CC BY-NC-SA 4.0},
  note        = {Open-source interactive intelligent textbook; 27 chapters,
                 37 MicroSims, 300 concepts, 301,873 words}
}

To cite a specific chapter, append the chapter title and URL — for example:

Pothula, A. (2026). Chapter 1: Introduction to Graphs and Networks. In Graph Neural Networks: An Intelligent Interactive Textbook. https://AnvithPothula.github.io/graph-neural-networks-textbook/chapters/01-intro-to-graphs/

License

This work is released under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (CC BY-NC-SA 4.0). You are free to share and adapt the material for non-commercial purposes as long as you give appropriate credit and share your adaptations under the same license.

References

  1. Ying, R., et al. (2018). Graph Convolutional Neural Networks for Web-Scale Recommender Systems. KDD 2018. https://dl.acm.org/doi/10.1145/3219819.3219890 ↩

  2. Jumper, J., et al. (2021). Highly accurate protein structure prediction with AlphaFold. Nature, 596, 583–589. https://www.nature.com/articles/s41586-021-03819-2 ↩

  3. Grand View Research. (2024). Knowledge Graph Market Size, Share & Trends Analysis Report. https://www.grandviewresearch.com/industry-analysis/knowledge-graph-market-report ↩

  4. Hu, W., et al. (2020). Open Graph Benchmark: Datasets for Machine Learning on Graphs. NeurIPS 2020. https://arxiv.org/abs/2005.00687 ↩

  5. Stokes, J. M., et al. (2020). A Deep Learning Approach to Antibiotic Discovery. Cell, 180(4), 688–702. https://doi.org/10.1016/j.cell.2020.01.021 ↩