Learning Graph: Graph Neural Networks

Open Learning Graph Viewer Fullscreen

This section contains the learning graph for the Graph Neural Networks textbook. A learning graph is a Directed Acyclic Graph (DAG) of concepts where every edge represents a learning dependency: concept A must be understood before concept B.

The graph has 300 concepts and 626 dependency edges organized across 15 taxonomy categories — from foundational math prerequisites at the left, through core GNN architectures in the middle, to frontier topics like LLM+GNN integration and graph foundation models at the right.

How to Read the Graph

• Left side (roots): Foundational concepts with no prerequisites — matrix multiplication, gradient descent, the undirected graph.
• Right side (leaves): Advanced terminal concepts — ULTRA, OFA, RelGNN, DiGress.
• Arrow direction: An arrow from A to B means "understand A before B."
• Node color: Encodes taxonomy category (see legend below).

Files in This Section

File	Description
Concept List	Numbered list of all 300 concepts
Learning Graph CSV	Full dependency graph with taxonomy labels
Learning Graph JSON	vis-network format for interactive viewer
Concept Taxonomy	15-category taxonomy with color assignments
Quality Metrics	DAG validation, indegree analysis, chain lengths
Taxonomy Distribution	Concept counts per category
Course Description Assessment	97/100 quality report

Taxonomy Color Legend

Color	TaxonomyID	Category	Count
SteelBlue	PREREQ	Prerequisites	20
DodgerBlue	FOUND	Graph Fundamentals	37
Teal	ALGO	Classical Graph Algorithms	33
DarkSlateBlue	EMB	Node Embeddings	17
Indigo	GNN	GNN Architecture	44
MediumPurple	THEORY	GNN Theory	17
DarkOrchid	TRANS	Graph Transformers	12
Crimson	KG	Knowledge Graphs	25
DarkRed	HETERO	Heterogeneous Graphs	8
DarkGreen	APP	Applications	23
OliveDrab	SCALE	Scalability	10
DeepPink	GEN	Generative Models	7
DarkGoldenrod	ADV	Advanced Topics	18
Orange	TRAIN	Training & Optimization	20
DimGray	TOOLS	Tools & Frameworks	9