Project Descriptions (2021-22)

The formal project descriptions are at this site

Masters Student Selection Process

As you have probably noticed, the AI Master's program here at IDI has become quite competitive: not every student who wants to do AI has that opportunity. Much of this problem stems from limited total thesis-advising capacity within the Data and Artificial Intelligence (DART) section. Unfortunately, this problem resurfaces when we professors choose students for projects: we cannot always take them all, and in some years an individual professor must reject more students than (s)he accepts. Today, there is no standard policy across IDI (or DART) for how professors manage these project requests, although we all use DAIM to post our projects and to accept/reject student requests. In the past few years, I have tried to standardize my own approach to the process, and what follows is hopefully useful information for any student who applies to any of my projects.

I try to get MOST of my projects posted on DAIM by April 15th, but a few extra may pop up in late April or early May.
I do NOT operate on a "first come, first serve" basis. Signing up early for one of my projects gives no extra priority over signing up in mid May.
However, at some (undisclosed) later date, usually somewhere between May 15th and June 1, I go through all of the requests and make my accept/reject decisions. There is usually a little "back and forth" for a few hours (or days) as accepted students accept or reject me, but this does not extend into the summer. A few days after I start this process, it ends and my students and projects are finalized for the year.
Just because a project is on my DAIM page does NOT mean that I will accept at least one student for that project. The decision of which students to accept and reject can be very complicated, and sometimes, a project falls out even though students have requested it.
Projects that are "in cooperation" with an industrial partner (or another NTNU department) are given a high priority, but once again, I cannot guarantee support for that project until I have seen the full list of applicants to my projects. For cooperating partners, cross-posting with several professors on DAIM is often wise. Although cooperative projects tend to be less work for the "internal" NTNU professor, there are still important responsibilities attached to any advisory role, including reading drafts, providing general advice, attending meetings with the partner(s), etc.
Students occasionally propose their own project ideas to me, but I am less likely to accept these than I was several years ago when the AI advising loads were much lighter. Still, a student-proposed project (that I agree to) must appear on DAIM, giving many students an opportunity to sign up for it. The student(s) proposing the project will have first priority in that case, assuming that I eventually agree to advise the project.
In most cases, I do NOT consider a student's grades in making my selections. The exceptions are a) some industrial projects, where the corporate partner needs a "top" student, and b) some of my personal projects that I may consider "very difficult". Such projects are not marked in any special way on DAIM, but at some point, I may realize that a grade check is in order and request grades from the corresponding students.
Any corporate partner who wants "only top" students, must explicitly request this; and all students applying to that project will be required to send their complete CV and NTNU transcript (with all grades) to the corporate partner.
I have no strict prioritisation criteria, but, in general, students who give a high priority number (1 or 2) to one of my projects are more likely to get it than those giving a lower number. That's pretty obvious.
One key prioritisation is for computer-science students who have specialized in AI. Students from other disciplines are rarely chosen due to the large number of AI students who I feel that we AI professors need to be advising.