1. Define a one-term project in the domain of artificial intelligence, either yourself, or with help of your supervisor;
2. Study literature, get background knowledge;
3. Solve the problem, possibly creating a non-trivial piece of software;
4. Present your solution: (a) orally during the final seminar, (b) in writing in a scientific report, and (c) as code stored in the course repository.
The concept of the project course is based on the idea that you learn much during practical work on a project building upon knowledge from the area of intelligent systems (or any area, for that matter. But we want you to learn AI:-) The amount of work expected from a project group is 2 x 200 hours, so a substantial system might be created with this effort.
Organization and Location
The project takes place in the 4th LP. There is no dedicated location for it. The participants will work on the machines in the basement or on their own machines. The duration of time spent on the project should be of about four to five weeks full time. Each participant can work alone or collaborate with one other person, although we prefer groups of two people working together.
Introductory Meeting 21/3
Admin and general info by Jacek (firstname.lastname@example.org)
Behavioural Neuroscience Laboratory by Andreas Heuer (email@example.com)
AI in Medical and Sustainability Research by Sonja Aits (firstname.lastname@example.org)
Autonomous Driving Systems (pages 10-11 of the intro) by Song Qunying (email@example.com)
Digital wall assessment of repointing needs by Maike Klöckner (firstname.lastname@example.org)
Robotics and RL by Matthias Mayr (email@example.com)
Learning for Robotics by Alexander Dürr (firstname.lastname@example.org)
Language processing by Pierre Nugues (email@example.com)
Two more possible supervisors available for discussion of ML-related projects are Erik Hellsten (firstname.lastname@example.org) and Marcus Klang (email@example.com).
We have the final seminar scheduled on May 25th, between 10:15 and 15:00, in E:2116. Each group will some 12 minutes (plus up to 3 minutes for discussion) to present their results, so please plan your presentation carefully: what problem have you tried to solve, with what method, what are the results, what is your conclusion. Make the presentation as attractive as you personally would like someone else's presentation to be. You may use the provided LU PowerPoint format file (choose 4:3 or 16:9 ratio), or something similar. Please make sure your presentation is in English, as some of the course participants do not know Swedish. Please read the presentation guidelines here before you give your talk: here [local copy] or Simon Peyton Jones' How to give a good research talk.
In the presentation, you will shortly describe the background, your system (architecture and outline of your algorithms), and results. You should provide some kind of evaluation and ideally show a demonstration.
Report and Programs
After your presentation, you will write a report of 4 to 8 pages that you will hand in together with your slides, and your programs (possibly through a public versioning repository). You will write your report as a research paper, like the ones you probably read when carrying your project out.
Please, use Latex and the IEEE or Association for Computational Linguistics styles to compose your report so that we have a uniform presentation across all the papers. The styles are available here: www.ctan.org/pkg/ieeetran/ and acl2020.org/authors/camera-ready-instructions/. Please use the A4 page size. Use also the Latex/Bibtex tool for your references. Should you have questions about it, please ask us. Görel Hedin wrote useful guidelines on how to write a report, that you can read here.
When you are done with your project, please send to your supervisor and to Jacek:
1. the final report in PDF with the Latex sources. Do not paginate it. Please state whether it is OK to publish it on the course web page;
2. your presentation slides in PDF (preferably), Powerpoint, OpenOffice, or similar formats. Please state also whether you agree to publish them on the course web (see earlier years in the left menu);
3. the code in a zipped archive, alternatively a link to a repository with it.
The deadline to hand in the report, the slides, and the code is June 8th, 2022, 23:59 CET.