Course syllabus

Course is offered by the department of Computer Science and Engineering

Contact details

Lecturer and examiner: Simon Olsson (simonols@chalmers.se) -- English only please.

Teaching assistant: Selma Moqvist <mselma@chalmers.se>

In-person only and there will be no recording of in-class sessions.

Course purpose

The purpose of this course is to give a broad introduction of structured machine learning. Structured machine learning involves using knowledge about the data domain and the data generating process to formulate learning problems which are more data efficient, generalize better, and hopefully also scale to larger problems.

The course will cover the following three broad themes:

• Data Generating Processes --- How is our data generated and what do we know about it and the resulting domain of our data?
• Geometric Deep learning --- How can we impose useful restrictions on the models that we learn that are in line with what we know about our data?
• Generative models --- How can we mimic the data generating distribution?

Schedule

Tuesdays 8.00-11:45 (Lectures)

Wednesdays 8.00-9:45 (Tutorials/labs)

Fridays 8.00-11:45 (Lectures)

Do not get scared off by the long lectures, they are broken up into smaller sessions with question and in-class problem solving.

 

Lecture overview

	Day	Date	Time		Room
w36	Tuesday	2026-09-01	08:00	11:45	EL52	Introduction, course overview
w36	Friday	2026-09-04	08:00	11:45	EL52	Data generating processes, high-dimensional data
w37	Tuesday	2026-09-08	08:00	11:45	ML13	Geometric priors
w37	Friday	2026-09-11	08:00	11:45	MB	Grids
w38	Tuesday	2026-09-15	08:00	11:45	MB	Graphs and sets
w38	Friday	2026-09-18	08:00	11:45	FL61	Groups and Convolutions
w39	Tuesday	2026-09-22	08:00	11:45	EL53	Convolutions and group representations
w39	Friday	2026-09-25	08:00	11:45	EL52	Application domains: Molecules
w40	Tuesday	2026-09-29	08:00	11:45	EL53	Generative Models 1: Ambient and latent spaces, autoregressive, bayesian networks
w40	Friday	2026-10-02	08:00	11:45	EL51	Generative Models 2: Data manifold, denoising, Variational Auto Encoders
w41	Tuesday	2026-10-06	08:00	11:45	EL53	Generative Models 3: Normalizing Flows
w42	Tuesday	2026-10-13	08:00	11:45	EL53	Generative Models 4: EBMs and Diffusion models
w42	Friday	2026-10-16	08:00	11:45	EL53	No lecture --- Self-Study for Projects proposals
w43	Tuesday	2026-10-20	08:00	11:45	EL53	The Bitter Lesson; its impact and where do we go from here
w43	Friday	2026-10-23	08:00	11:45	EL53	Project proposal presentations

Tutorial schedule

Week	Date	Start time	End time	Room
w36	2026-09-02	08:00	09:45	EL52
w37	2026-09-09	08:00	09:45	EL52
w38	2026-09-16	08:00	09:45	EL52
w39	2026-09-23	08:00	09:45	EL52
w40	2026-09-30	08:00	09:45	EL52
w41	2026-10-07	08:00	09:45	EL52
w42	2026-10-14	08:00	09:45	EL52
w43	2026-10-21	08:00	09:45	EL52

TimeEdit

Student representatives

MPCAS   simon02eriksson@hotmail.se      Simon Eriksson
MPCAS   redpowertop@163.com     Kaige Guan
MPDSC   leijn933@gmail.com      Jianing Lei
MPDSC   soosaki001@gmail.com    Ákos Soós
MPDSC   monicaruth202@gmail.com Monica Ruth Fabiola Winata

 

Assignment Schedule PRELIMINARY

	Release	Deadline	Final re-submission opportunity
Hand-in 1	01-Sep	08-Sep	NA
Hand-in 2	08-Sep	15-Sep	NA
Hand-in 3	15-Sep	22-Sep	NA
Project 1	21-Sep	28-Sep	5-Oct
Project 2	28-Sep	5-Oct	13-Oct
Project 3	5-Oct	13-Oct	20-Oct
Essay	2-Oct	19-Oct	NA
Essay review	19-Oct	22-Oct	NA

Project discussion time

The discussion about one of your projects will happen on Oct 26, in EL41 between 9-16. More details will follow.

Course literature (to be updated)

Large parts of the course material is loose adaptations from Geometric Deep Learning summer schools.

Primary reference:

Bronstein, Bruna, Cohen, and Veličković:  Geometric deep learning. (Free proto-book available: https://arxiv.org/abs/2104.13478)

Selected primary literature and lecture notes TBA

Extra literature:

Serre "Linear Representations of Finite Groups" (1977) https://link.springer.com/book/10.1007/978-1-4684-9458-7

Background references for repetition:

Deisenroth, Faisal, and Ong "Mathematics for Machine Learning" (2021) https://mml-book.github.io/book/mml-book.pdf

Shapira "Linear Algebra and Group Theory for Physicists and Engineers." (2019)  https://link.springer.com/book/10.1007/978-3-030-17856-7

Petersen and Pedersen "The Matrix Cookbook" https://www.math.uwaterloo.ca/~hwolkowi/matrixcookbook.pdf

Git crash course: https://julienpascal.github.io/slides/intro_git/#/

Further math background Mathematical Foundations of Geometric Deep Learning

Practical extra information:

37 reasons why you neural network is not working (Blog post)

 

Course design

The course is based on four components

• Self study: reading and video lectures (where applicable)
  
• Interactive hybrid lectures
• Individual project work
• Peer-assessment

and follows a blended classroom structure.

Self study:

Before each week reading and other material is provided for preparation. Preparation is key for the success of the interactive lectures. In these lectures, we will be moving through concepts and problems, and solve them in teams. There will not be repetition of materials given in advance, we assume attendees have read and followed provided material in advance. The first lecture is an exception.

Lectures:

The in-class lectures are interactive and focus on problem solving in teams, interspersed with micro lectures, and discussions (tuesdays and fridays). Attendance is strongly advised.

Tutorials:

Tutorials are lead by TA Selma Moqvist.  Selma will be available to help with assignments and project work. Further she will provide give consolidated feedback on assignments and walk through some of the solutions. Finally, during the project assignment weeks she will do small tutorials and walk-throughs on some of the practical implementations you are asked to do. 

Learning objectives and syllabus

Study portal

Examination and assessment

The examination consists of submissions, homework assignments, peer assessment and reports, as specified in the course syllabus. To pass the course, the following elements must be completed by the end of the course:

• three take-home assignments (pass/fail) — at least two must be passed,
• three practical projects (submission requirements below),
• one oral project defense (graded),
• one project proposal for a 6-month research project, max 1500 words (pass/fail), presented as a 5-minute pitch (graded),
• peer review of project proposals, including acting as opponent at one presentation (graded).

Your final course grade is determined by the three graded components — the oral defense, your work as peer reviewer, and your proposal presentation — as described under Final grade below.

Take-home assignments

Three take-home assignments are given during the course and graded pass/fail. You must pass at least two of the three. Feedback is given as a consolidated walkthrough of common mistakes and good solutions in the tutorial sessions; no individual written feedback is provided.

Practical projects

The course has three individual practical projects with deadlines given in the table above. For each project you must:

• complete the code and pass the provided unit tests,
• submit a written report of at most 4 pages (excluding references),
• include a statement on the use of aids (see Use of aids below).

Submissions that do not meet these requirements are returned for completion. All three projects must meet the requirements before your oral defense can take place. Feedback on projects is given as a consolidated summary to the whole class after each deadline.

Oral defense

After the final project deadline, every student attends one individual oral defense of about 10 minutes. At the start of your slot, one of your three projects is drawn at random. The defense centres on the drawn project, but questions may concern any of your submitted work. Bring a computer on which you can run your code and unit tests.

The defense is graded as follows:

• Insufficient — you cannot account for the work you submitted.
• 3 — you can clearly account for what was done and why: you can explain your code, your design choices and your results, and your unit tests pass.
• 4 — in addition, you demonstrate sound trade-off analysis, for example comparing your chosen approach against alternatives in terms of expressivity, computational cost or sampling complexity.
• 5 — in addition, you demonstrate deep conceptual understanding of the course material: you can connect your project to the underlying theory and reason convincingly about extensions and failure modes.

Project proposal and presentation

You write a proposal for a 6-month research project building on the course themes, max 1500 words. The written proposal is graded pass/fail; passing requires a clearly stated problem, a proposed method grounded in the course content, and appropriate references.

The proposal is assessed through the written document together with its oral presentation: a 5-minute pitch followed by questions from your opponent and the audience. The presentation is graded:

• Insufficient — the pitch does not communicate the proposal, or the questions reveal that you cannot account for it.
• 3 — clear problem statement and method, grounded in course concepts, with reasonable answers to questions.
• 4 — in addition, convincing treatment of feasibility and of the trade-offs in the proposed design.
• 5 — in addition, an original and compelling proposal, defended with depth under opposition.

Peer review

During an in-class review session you write structured reviews of two of your peers' proposals. Your structured reviews (max 1 page per review) must be submitted before the presentation. Further you will need to act as opponent at their presentation. Your work as reviewer is graded:

• Insufficient — reviews are missing or perfunctory.
• 3 — accurate summaries (Written only) and relevant, respectful critique of the reviewed proposals. (Written and Oral)
• 4 — in addition, you identify substantive strengths and weaknesses using concepts from the course. (Written and Oral)
• 5 — in addition, you contribute non-obvious insights and concrete suggestions that would clearly improve the proposed research. (Written and Oral)

Final grade

To pass the course you must complete all elements listed at the top of this page and receive at least grade 3 on each of the three graded components. No graded component can compensate for another.

For students who pass, the final course grade is given by the sum of the three component grades:

Sum of component grades	Final grade
9–11	3
12–13	4
14–15	5

For example: 3 + 4 + 4 = 11 gives final grade 3; 3 + 4 + 5 = 12 gives final grade 4; 4 + 5 + 5 = 14 gives final grade 5.

Late submissions

For each started day a project report or the proposal is submitted late, 0.5 points are deducted from your point sum above. Deductions can lower your final grade, but cannot cause a failing grade if all graded components are at least 3. Submissions more than 7 days late are not accepted at the ordinary examination occasion; you will instead be referred to the re-examination period, and your defense or presentation is scheduled there. Remember, exceptions are possible in exceptional cases; if you think you have one, please reach out.

Re-examination

• Oral defense (insufficient or missed): a new defense with a new random draw is offered in the re-examination period.
• Presentation (insufficient or missed): a make-up presentation slot is offered in the re-examination period.
• Peer review (insufficient or missed): an alternative individual review assignment is given in the re-examination period.
• Take-home assignments and projects that are not passed can be completed at the next examination occasion.

Use of aids

All aids are allowed throughout the course — generative AI models (ChatGPT, Gemini, Claude, etc), discussions with friends, Stack Overflow, and so on. However, your use of aids must be thoroughly documented, and you must be able to account for all work you submit. However, I do not accept the use of AI models during class.

Every project report and the proposal must include a use statement detailing:

1. which aids you used and how,
2. how you verified their output,
3. whether you judge that they saved you time,
4. the most consequential error you caught in an aid's output, and how you caught it.

The use statement is part of the pass requirements for each report.