Machine Learning

The first course in the machine-learning chain of the faculty, and the foundation for everything the lab works on. It starts with the question of what a learning task actually is, and ends with students training their own deep networks on real data.

What you learn

• Recognise machine-learning tasks and sort them into the right category: supervised, unsupervised, reinforcement.
• The main methods of supervised and unsupervised learning, and when each of them is the right choice.
• Read the topology of a deep neural network, design your own model, implement it and train it.
• Work through scientific papers and complex topics on your own, and judge what a result is worth.
• Write your own software for a practical problem instead of running someone else’s notebook.

How it is taught

Lectures alternate with lab sessions in which the methods are implemented in Python. The examples come from the lab’s own field: images and point clouds of street scenes, pedestrians and cyclists.

What comes after it

Computer Vision and Autonomous Driving build directly on this course. Students who enjoy it often continue with a project or a thesis in the lab, or work there as student assistants.