After successfully passing the Embedded AI course, a student can:
- explain basic ML/AI algorithms
- construct ML/AI algorithms using basic building blocks
- analyse functional and non-functional requirements of AI applications and based on that, select the most suitable embedded processing platform
- implement (building blocks of) ML/AI algorithms onto embedded processing platforms
- analyse the effects of implementation on the performance of the algorithm
- assess the overall quality of the implementation (link between algorithm and implementation)
In general terms, the Embedded AI course consists of the following topics:
Applications (computer vision, natural language processing, localization, health monitoring)
- Basics of machine learning and inference
- Elementary building blocks of AI/ML algorithms
- Non-standard computing- and resource-bounded (time, space, power) platforms for Embedded AI (microcontrollers, FPGA’s) Communication and sensing
- Mapping of algorithmic building blocks onto computing platforms.
'< 10 students' => oral examination
- Test on the contents taught during the lectures (50%).
'> 10 students' => written examination
Project groups of 3-5 students, depending on the number of participating students.
- Presentation and demonstration of project results + Written report (50%)