Course Information Fall 2019

CS224W: Machine Learning with Graphs

Instructor Jure Leskovec

Co-Instructor Michele Catasta

Office Hours: exclusively about projects, on appointment (arranged via personal email)
Lectures 3:00-4:20pm Tuesday and Thursday in NVIDIA Auditorium, Huang Engineering Center
Course website http://cs224w.stanford.edu

Contact
• For course related questions, use Piazza: http://piazza.com/stanford/fall2019/cs224w
(use your @stanford.edu address to register)
• For personal matters, or in emergencies, e-mail us at [email protected]
• SCPD students can attend office hours remotely via a Google Hangout; the link will be posted
on Piazza just before the office hours start.
TAs (Office Hours in Huang Basement, check website for the schedule)
• Christina Yuan [head TA]
• Lingzi (Liz) Guo
• Benjamin (Ben) Hannel
• Kuangcong (Cecilia) Liu
• Vasco Portilheiro
• Andrew Wang
• Alexis Goh Weiying
• Zhitao (Rex) Ying

Topics
• Six degrees of separation • Influence maximization in networks
• Models of the Small World • Link analysis, PageRank
• Network Motifs, Structural Roles in networks • Knowledge Graphs and Metapaths
• Message passing and Node classification • Finding communities and clusters in networks
• Node Representation Learning, Node2Vec • Spectral clustering and large scale community
• Graph Neural Networks structure in networks
• Power-laws and Preferential attachment • Modularity and large scale community struc-
• Models of network evolution ture in networks
• Cascading behavior in networks • Network Inference
• Models of network cascades • Network Robustness

Course Grading
• 1 setup homework (1%) + 3 (9.6% each) homework requiring coding and theory (30%)
• 2 hour open-book exam (30%)
• Final project: proposal, milestone report, poster presentation, and final report (40%)
• Piazza and course participation, contributions to the SNAP codebase (extra credit factored in during
grading)
 CS224W: Machine Learning with Graphs, http://cs224w.stanford.edu 2

Assignment Policy
Questions We try very hard to make questions unambiguous, but some ambiguities may remain. Ask (i.e.,
post a question on Piazza) if confused or state your assumptions explicitly. Reasonable assumptions
will be accepted in case of ambiguous questions.
Honor Code We strongly encourage students to form study groups. Students may discuss and work on
homework problems in groups. However, each student must write down the solutions independently.
That is, each student must understand the solution well enough in order to reconstruct it by him/herself.
In addition, each student should write down the set of people whom s/he collaborated with.
Late Submissions Each student will have a total of 2 late periods to use for homework. Homework are
due on Thursdays (11:59pm PT) and late periods extend to midnight (11:59pm PT) of the following
Monday. Only one late period can be used per assignment, and no assignment will be accepted more
than one late period after its due date (modulo a 15min grace period). Late periods can also be used
for the submissions related to the final project (i.e., proposal and milestone) except the final writeup.
Assignment Submission All students (SCPD and non-SCPD) submit their homework via Gradescope
(http://www.gradescope.com). Students can typeset or scan their homework. Make sure that you
answer each question on a separate page. That is, one answer per page regardless of the answer length.
You will be added to Gradescope as soon as you officially enroll to CS224W. When Gradescope requires
you to upload code for a specific question, include all the code pertaining to that single question into
a compressed file and upload it.

Prerequisites
Students are expected to have the following background (recitation sessions will refresh these topics):
• Knowledge of basic computer science principles, sufficient to write a non-trivial computer program
(e.g., CS107 or CS145 or equivalent are recommended.)
• Familiarity with the basics of probability theory (CS109 or Stat116 are sufficient but not necessary.)
• Familiarity with the basics of linear algebra (any one of Math 51, Math 103, Math 113, or CS 205
would be much more than necessary.)

Materials
Notes and reading assignments will be posted on the course website. Selected readings for the class will be
from: Networks, Crowds, and Markets: Reasoning About a Highly Connected World by D. Easley and J.
Kleinberg (PDFs available online).

Important Dates
Assignment Out Due
Homework 0 now October 3
Homework 1 September 26 October 10
Project proposal October 17
Homework 2 October 10 October 24
Project milestone November 7
Homework 3 October 24 November 14
Exam November 19
Project final report December 10 (no late periods)
Project poster session December 12, 12:15-3:15pm

We will also hold 2 review sessions (which will be recorded) in the first two weeks of the course:
• SNAP.PY and Google Cloud tutorial: Skilling Auditorium, Friday 09/27, 3:00-4:20pm
• Review of probability, linear algebra, and proof techniques: Skilling Auditorium, Friday 10/4, 3:00-
4:20pm