Dear all

What do Mesopotamian artifacts, Tokamak legs, Histology, Gravitational lenses, Applause Recognition, Rooftop Solar, Stroke Rehabilitation, Air Quality Prediction, Lego Piece Selection, Invisibility Cloaks, Twitter Astroturfing, Drone Navigation, Classical Music and Earthquakes have in common?
In the five editions so far, 528 collaborative projects have been successfully completed, across 105 different labs of EPFL and other institutions, as part of the student projects in the machine learning course CS-433
https://www.epfl.ch/labs/mlo/machine-learning-cs-433/

This year we are running this collaboration initiative again across campus, trying to connect you as a hosting lab to 550 very talented and enthusiastic master students in the course.

During 2 months from November to end of December, they will do their main course project in groups of 3 - part-time.

If your lab agrees, a group of 3 of our students can contribute to a real-world research application in your lab - ANY lab of an academic* institution, from any diverse area of science. Do you have an interesting dataset and question available for a short interdisciplinary project, exploring ML methods for a new application? See this website https://www.epfl.ch/labs/mlo/ml4science/ for a list of previous topics we hosted in the last years. (*typically Swiss academic institutions, international institutions and NGOs might qualify as well - please contact us)

If your lab is interested, what we need from you is not much for now - just a contact address and the fact that your lab is considering hosting a group or several, so students can get in touch with you. Please fill this form (right here) to express your interest.
It could be good to already think of available datasets, suggested prediction tasks or other scientific questions, and potential evaluation metrics.

Later by November 10th, we will ask for a definitive commitment from the head of your lab, if you are willing to host a particular group(s) of your choice, or not. We will use a different google form for that later. The matching process works by students contacting you, based on their interest in the preliminary proposed topic. After you talk with the students, you can decide to commit or not to a group (before Nov 10th). If you decide to host a particular group, the professor in the end needs to give us a grade suggestion for the domain-specific merits of the finished project (all projects will be due December 21st). The students will submit a 4 page PDF report, and the code. We will take care of the submission system, and our assistants will help assess the technical parts of each project, but we need your help for assessing the domain specific merits of the project you host. This project counts 30% to their ML course grade. The volume of work corresponds to about 2 to 2.5 credit points.

Please feel free to share this call further.
This year we'd particularly encourage project ideas with an ethics or society angle - but there are no restrictions on topics, and students chose freely.

We are really looking forward to as many new interdisciplinary projects as possible!


martin jaggi & nicolas flammarion
IC