Welcome to UAI

Welcome to our course on Uncertainty in AI!

Below we sketch the practical arrangements for the UAI course in 2023-24. 

  • LECTURES: We are following the Flipped Classroom principle. That means that all lectures (covering the theory) available in video’s and that you have to watch these from home before the lecture. Each week we have a live interactive sessions (see timetable) with a particular topic (see course documents). All questions asked before Thursday 12:00 in the nextbook platform will be tackled in the live session. In the live session some additional elements such as quizzes can be done. The live sessions will not be livestreamed nor recorded due to the interactive nature.
    However, for students not able to attend the live sessions, written responses to the questions in nextbook will be provided and if possible the quizzes used during the interactive session will be shared on Toledo.
  • EXERCISE SESSIONS: Are collaborative sessions supported by teaching assistants. All solutions together with video’s explaining the solutions step by step will be made available online. There are 6 Exercise sessions; one session may be hands-on.
    Students are divided in two groups for the exercise sessions based on your last name:
  • Group A: all students with surnames starting with A to M
  • **Group B: all students with surnames starting with N to Z
    **Only if you have overlap with other courses you should change groups. If there is a capacity problem, let us know!
    Use the discussion board to ask questions related to the exercise sessions.
  • ASSIGNMENTS: the course has biweekly assignments -> see post below. These bi-weekly assignments will be published under “Assignments” in the left menu..

We wish you a lot of success!

Tinne & Luc

Lectures

All lecture material (videos and slides) are offered online in a learning module (see next).
Additionally, but optionally, we offer weekly Q&A sesssion at Friday 9:00 (see course schedule) where we handle questions regarding the material (see below) and potentially solve some exercises interactively.

Beware the schedule: each Q&A is devoted to a particular topic. Questions regarding this topic should be asked before Thursday 12:00 preceeding the Q&A session using the nextbook platform (see below).

29/09 Lecture 1: Introduction and probabilities (Tinne)

06/10 Lecture 2: Introduction to graphs and Bayesian Networks (Tinne)

13/10 Lecture 3: Bayesian Networks, Conditional independence, d-separation (Luc)

20/10 Lecture 4: Markov Nets, (in)dependence maps (Luc)

27/10 Lecture 5: Inference I – graphical models to answer queries (Tinne)

03/11 No Lecture

10/11 Lecture 6: Inference II (Luc)

17/11 Lecture 7: Learning from fully observable data (Luc)

24/11 Lecture 8: Learning from partially observable data (Tinne)

01/12 Lecture 9: Logic and probability (Luc)

08/12 Lecture 10: Approximate inference and sampling (Tinne)

15/12 Lecture 11: Dynamic models (Tinne)

22/12 backup

Assignments

Dear students, Friday October 6 we will post the first assignments.
Time to repeat some arrangement around the assignments, which you could already find in the slides of the first lecture.

Goal? The goal of the assignments is to keep you on track with the course, and to reward you for staying on track.

Topic? The assignments will cover the topic of the two lectures. Remark that therefore, at the time the assignment is posted you will possibly not have seen all required topics. The upcoming lecture will handle the missing topics.

Evaluation? The assignments are part of the permanent evaluation of the course and count for 10% of the total score (2 of the 20 points). You have to hand in at least 3 assignments (of the six published). For the evaluation part we will just take into account your 3 best assignments if you submit more than 3. Beware: the probability theory test, which was just for self-positioning, does not count as an assignment.

Feedback? We will publish a model solution for the assignment shortly after the deadline. Once your assignments are graded you can find the grades in the grade center of the Toledo page of this course. Like this you can keep track of your grades. Remember, the assignments are there to support you to stay on track with the course.

When published? The assignments are published on Toledo, bi-weekly after lecture (starting after second lecture)

Deadline? 10 days after publication, i.e. the Monday one week later at 16:00.

How to submit? Submit electronically on Toledo, in the assignment itself! We do not accept emails nor uploads in other parts of Toledo.

What to submit? We prefer hand-written and scanned (do not lose time by typesetting, etc.) solutions. Clearly indicate the final answers (e.g. box it)! It is required to fill in the answer sheet, which allow our Teaching Assistant to spot correct answers fast.

Is it a group work? No. The assignments should be individual work!
We do check for plagiarism.

Can I redo the assignments in summer if I fail the course? No. As the assignments are a key part of the permanent evaluation, which allow you to stay on track with the course, you cannot resit them.

Good luck!

Exam

  • The exam is closed book but you can use the formularium. Bring a printed copy. Make sure you did not add any notes yourself.
  • Bring a simple calculator. Make sure the memory of the calculator is empty,
  • You will have 3 hours to complete the exam.