Schedule

Lectures are on Tuesdays, periods 2–3 (10:00–11:50) and practice sessions are on Thursdays, period 3 (11:00–11:50). The materials for each week are listed below the timetable.

Weekly Timetable

Mon

Tue

Wed

Thu

Fri

109:00

210:00

311:00

412:00

513:00

614:00

715:00

816:00

Lecture 10:00–11:50 Periods 2–3 · Sample code

Practice 11:00–11:50 Period 3 · Practice code

Week by Week

Week 1 · Aug 31 – Sep 4

Introduction to Graphs and Node Centrality

Graph basics with NetworkX and node centrality measures: degree, betweenness, closeness, eigenvector, Katz, and PageRank.

Lecture Tue, Sep 1 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Sep 3 · 11:00–11:50 practice practice (Colab)
Assignment Due Wed, Sep 9 · 23:59 Assignment #1: Graph Basics and Node Embeddings solutions

Week 2 · Sep 7 – Sep 11

Message Passing and Graph Neural Networks

Relational classification and label propagation, the message passing framework, and GCN on Zachary's Karate Club.

Lecture Tue, Sep 8 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Sep 10 · 11:00–11:50 practice practice (Colab)
Assignment Due Wed, Sep 16 · 23:59 Assignment #2

Week 3 · Sep 14 – Sep 18

The Expressive Power of Graph Neural Networks

Computational graphs, color refinement and the 1-WL test, GIN, and architectures beyond 1-WL.

Lecture Tue, Sep 15 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Sep 17 · 11:00–11:50 practice practice (Colab)
Assignment Due Wed, Sep 23 · 23:59 Assignment #3

Week 4 · Sep 21 – Sep 25

Towards Scalable Graph Neural Networks

Mini-batch training on large graphs: GraphSAGE, FastGCN, AS-GCN, GraphSAINT, and ClusterGCN.

Lecture Tue, Sep 22 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Sep 24 · 11:00–11:50 practice practice (Colab)
Assignment Due Wed, Sep 30 · 23:59 Assignment #4

Week 5 · Sep 28 – Oct 2

Deeper GNNs: Over-smoothing, Over-squashing, and Under-reaching

Why GNNs cannot simply go deeper, and remedies such as skip connections, normalization, JKNet, and curvature-based rewiring.

Lecture Tue, Sep 29 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Oct 1 · 11:00–11:50 practice practice (Colab)
Assignment Due Wed, Oct 7 · 23:59 Assignment #5

Week 6 · Oct 5 – Oct 9

Heterophily and Graph Structure Learning

Homophily vs. heterophily and heterophily-aware GNNs (H2GCN, GPR-GNN); graph structure learning that jointly or adaptively learns graph topology and node representations.

Lecture Tue, Oct 6 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Oct 8 · 11:00–11:50 practice practice (Colab)

Week 7 · Oct 12 – Oct 16

Attentive GNNs and Graph Pooling

Graph Attention Networks (GAT) and attention in heterogeneous graphs; hierarchical graph pooling (DiffPool, SAGPool).

Lecture Tue, Oct 13 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Oct 15 · 11:00–11:50 practice practice (Colab)

Week 8 · Oct 19 – Oct 23

Mid-Term Exam

Details will be announced in class.

Week 9 · Oct 26 – Oct 30

Explainability, Generalizability and Robustness

Interpreting GNN predictions (GNNExplainer, PGExplainer); adversarial attacks on graph data; defenses and robust GNN architectures.

Lecture Tue, Oct 27 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Oct 29 · 11:00–11:50 practice practice (Colab)

Week 10 · Nov 2 – Nov 6

Graph Transformers

From message passing to self-attention; positional and structural encodings (PE/SE) in graphs; representative models (Graphormer, SAN).

Lecture Tue, Nov 3 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Nov 5 · 11:00–11:50 practice practice (Colab)

Week 11 · Nov 9 – Nov 13

Generative Graph Models and Diffusion

Variational autoencoders for graphs (VGAE); graph diffusion models for generating novel molecular structures or networks.

Lecture Tue, Nov 10 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Nov 12 · 11:00–11:50 practice practice (Colab)

Week 12 · Nov 16 – Nov 20

Spatio-temporal Graph Neural Networks (STGNN)

Spatial and temporal dependencies; applications in traffic forecasting, weather modeling, and action recognition.

Lecture Tue, Nov 17 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Nov 19 · 11:00–11:50 practice practice (Colab)

Week 13 · Nov 23 – Nov 27

GNNs for Recommendation and Molecular Structure Learning

User-item bipartite graphs, LightGCN and PinSage, RecSys evaluation metrics; molecules as graphs, E(n)-equivariant GNNs, AlphaFold and drug discovery.

Lecture Tue, Nov 24 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Nov 26 · 11:00–11:50 practice practice (Colab)

Week 14 · Nov 30 – Dec 4

Heterogeneous Graphs, Knowledge Graphs, and GraphRAG

Knowledge graph representation learning (TransE, RotatE); synergy with LLMs through Graph Retrieval-Augmented Generation (GraphRAG).

Lecture Tue, Dec 1 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Dec 3 · 11:00–11:50 practice practice (Colab)

Week 15 · Dec 7 – Dec 11

Course Consolidation

Review and consolidation of the course.

Lecture Tue, Dec 8 · 10:00–11:50 sample code sample code (Colab)
Practice Thu, Dec 10 · 11:00–11:50 practice practice (Colab)

Week 16 · Dec 14 – Dec 18

Final Exam

Details will be announced in class.