CS 7280: Network Science

Instructor: Constantine Dovrolis
Semester: Summer 2024
Overall Rating: 8.6/10

---

✅ Pros

• Comprehensive coverage of core network‑science concepts
• Clearly defined project requirements
• Well‑structured lectures

❌ Cons

• Quiz questions could be refined as some questions are ambiguous
• Grading can be inconsistent, minor issues may cost points depending on the TA

---

🕒 Time Commitment

Roughly 10 hours per assignment and 4 hours per week on notes, modules, and quizes.

---

✍️ Assignments

The course starts with some heavy math. If it feels overwhelming, you’re probably not alone. Try and work through the equations and build intuition around the ideas. You won’t be expected to memorize proofs or derive formulas.

• Five projects (one per module)
• Each project ships as a ZIP containing a Jupyter notebook and datasets
• Assignment instructions are typically in the form of “implement function X” or “run code in cells foo and bar to answer question Y”
• Instructions are generally clear, and projects are relevant to the material in the modules and reinforce learnings
• Grading can be inconsistent and varies by TA

---

📖 Quizzes

Every lesson is accompanied by a quiz, about seven multiple‑choice questions, that can be tricky, both from the content and the semi-frequent ambiguous wording.

---

📚 Course Content

Module 1 · Foundations of Network Science
Lessons 1-2 introduce the field and refresh essential graph‑theory concepts. You’ll learn what distinguishes complex networks from trivial ones, survey classic applications (brain networks, cascades, ML on graphs), and review core graph structures—directed/undirected graphs, paths, trees, adjacency matrices, flows, and bipartite representations.

Module 2 · Network Structure & the Small‑World Phenomenon
Lessons 3-5 analyze real‑world degree distributions, the “friendship paradox,” Erdős‑Rényi vs. scale‑free networks, clustering coefficients, path lengths, and the emergence of small‑world properties. Hands‑on examples explore superspreaders, robustness to failure, and motif statistics.

Module 3 · Centrality, Cores & Communities
Lessons 6-8 dive into node importance and meso‑scale structure. Topics include PageRank, eigenvector and betweenness centrality, k‑core decomposition, rich‑club effects, modularity metrics, Louvain and other community‑detection algorithms, plus advanced ideas such as overlapping and dynamic communities.

Module 4 · Network Dynamics & Spreading Processes
Lessons 9-11 model how information, diseases, and influence traverse networks. You’ll study SI/SIS/SIR epidemic models, cascade and threshold diffusion, game‑theoretic adoption, inverse percolation, decentralized search, synchronization (Kuramoto), and coevolutionary networks.

Module 5 · Modeling, Inference & Machine Learning on Graphs
Lessons 12-14 cover generative and statistical tools for networks: preferential attachment, configuration and hierarchical random‑graph models; sampling and topology inference; link prediction; and modern ML techniques—node embeddings, graph neural networks, temporal and interdependent graphs, and deep generative models.

---

💬 Class Participation & Interaction

Ed Discussion was quiet and office hours were not offered, but the lectures and project PDFs are usually clear enough.

---

💭 Final Thoughts

Overall I enjoyed this course. Though it leans more theoretical, the material seemed useful from a practitioner’s perspective.