DS683 - Graph Neural Networks (FALL · 2025)

• Description
• Teaching Team
• Logistics
• Materials
• Schedule

Description (DS683.001)

Graphs provide a natural framework for representing complex relationships between various objects. Graph Neural Networks (GNNs) have gained significant importance in both academic research and industrial applications. This course introduces GNNs and explores foundational concepts, algorithms, and diverse applications. Students will learn the fundamentals of graph theory, and key models, e.g., Graph Convolutional Networks (GCNs), Graph Attention Networks (GATs), advanced graph diffusion models, and integrations of GNNs with sequential models for temporal graph modeling. The course will cover practical applications across fields like social networks, bioinformatics, and finance, focusing on hands-on implementation and problem-solving. By the end, students will be skilled in designing and applying GNN models to real-world datasets.

Teaching Team

Logistics

• Time: Friday, 2:30 PM - 5:20 PM EST (FALL 2025）
• Location: CULM LEC 3
• Mode: Face-to-Face
• Discussions: Feel free to participate in different discussion forums on Canvas
• Quizzes & Assignments: submit and post on Canvas
• Prerequisites: CS675 or DS675 or instructor permission.
• Contact: For external inquiries, personal matters, or emergencies, please email the instructor or TA directly.

Materials

There is no required course textbook. The course will draw material from several sources, including the instructor’s own notes. Some open resources include:

• William L. Hamilton. Graph Representation Learning
• Michael M.Bronstein, Joan Bruna, Taco Cohen, Petar Veličković. Geometric Deep Learning Grids, Groups, Graphs, Geodesics, and Gauges
• Albert-László Barabási. Network Science
• David Easley and Jon Kleinberg. Networks, Crowds, and Markets: Reasoning About a Highly Connected World

Grading

• Quizzes [10%]: Weekly Canvas quizzes reinforcing the material of each module, will help you keep up with the most important theoretical concepts. These quizzes are not proctored.
• Class Participation [10%]: You are expected to attend classes and participate in classes by listening and understanding class contents and asking related questions. You are also expected to participate in weekly Canvas discussions prompted by the instructor, with meaningful questions and answers related to the week’s topics or assignments.
• Homeworks [30%]: Assignments will be given biweekly to give you an opportunity to apply course concepts for that week. Four homework assignments of equal grading weight.
• Mid-term Exam [15%]: In-person exam, 90 minutes. Students are expected to bring a fully charged laptop, as the exam will be on Canvas with LockDown browser. Each student is allowed to bring at most 5 pages of notes. In the event the exam has to take place online, Respondus Monitor will be used for proctoring.
• Project [35%]: The project will consist of three milestones, with weights [10%, 5%, 20%]. You will have opportunities to iterate and revise your work based on peer, TA, and instructor’s feedback.

Schedule

The lecture slides are available for download in the Google Drive folder, while all weekly quizzes and biweekly assignments can be accessed on Canvas.

Week	Date	Topic	Readings	Due Work
Week 1	9/5	Introduction and Course Overview	[B1] Ch 1 [B2] Ch 1 [B3] Ch 1
Week 2	9/12	Traditional Graph Analysis	[B1] Ch 2 [B2] Ch 2	Hwk 1 out
Week 3	9/19	Graph Embedding with Shallow Neural Networks	[B1] Ch 3	Project milestone 1
Week 4	9/26	Modern Graph Neural Networks (Spectral vs Spatial methods)	[B1] Ch 5	Hwk 1 due
Week 5	10/3	Expressivity Theory of GNNs	[B1] Ch 7
Week 6	10/10	GNNs (Oversmoothing and Oversquashing)	Papers	Hwk 2 out
Week 7	10/17	Graph Transformers	Papers	Project milestone #2
Week 8	10/24	Graph ViT/MLPMixer + Midterm Review	Paper	Hwk 2 due
Week 9	10/31	Midterm Exam	Papers	Summary quiz
Week 10	11/7	Understanding Eigenvectors, Frequencies, and Smoothing		Hwk3 out
Week 11	11/14	Knowledge Graph Reasoning	[B3] Ch 4
Week 12	11/21	Advanced Knowledge Graph Reasoning: From GNNs to GraphRAGs and Foundation Models	[B1] Ch 9 Papers	Hwk 3 due
Week 13	11/26 (Friday Class Meet)	New Frontiers in GNNs (graph diffusion models, physics-informed GNNs, etc.)	Papers
Week 14	12/5	Project Presentation		Project milestone #3

* The schedule is subject to change. Please check Canvas for the most up-to-date information.

For more detailed course information, please refer to the syllabus and the Canvas modules!