The MachineLearnAthon is an Erasmus+ funded education research project which seeks to introduce a new teaching format centered around machine learning (ML) competitions.
Rather than starting with theory and only later moving to implementation, we use ML challenges as a roadmap for teaching: Given a real world prediction task, we create the material required to tackle it and present it to students in the form of micro-lectures (< 10 minutes) and tool introduction units. Students, working in (international and inter-disciplinary) teams can use the challenge data to directly apply the acquired knowledge to solve the posed challenge whenever and however they see fit.
We use public leader-boards to spur competition thus motivating the students to come up with creative approaches to improve their own initial solutions and outperform their peers.
The material is designed for teams of students from heterogeneous fields (Computer Science, Business Administration, Engineering and Logistics) with varying degrees of programming and ML expertise, from beginner, through intermediate to advanced.
The MachineLearnAthon is aimed at improving data literacy and hands-on ML knowledge (from data collection to deployment) and increase the awareness of both the potential and risk associated with ML. Furthermore, the international and inter-disciplinary setup of the project consortium serves to foster cooperation between both students and teachers.