Machine and Deep Learning

The range of problems that a computer scientist may encounter is very broad, from the very abstract (to determine if some problem can be solved by a computer) to very specific (to implement an electronic health record system). Due to the complexity and noisiness of the real world, a modern computer scientist has to rely more and more on the analysis and models of data. Machine Learning, and a recently successful extension of it, Deep Learning, deals with the analysis of data and the development of models.

To be able to apply these algorithms, the computer scientist needs to have a good understanding of the foundations of Machine Learning. This course provides a recapitulation of (un)supervised learning, classification, decision theory and overfitting. It introduces some classical statistical learning classifiers and discusses their complexity, and ways to adapt their complexity (regularisation). The course focuses further on Deep Learning, including backpropagation, Optimization, Convolutional Neural Networks, Recurrent Neural Networks, self-attention and Transformers.

Finally, the course discusses biases in datasets, and the design and analysis of Machine Learning experiments. This makes the student aware how the Machine and Deep Learning models can be applied in a responsible manner to real world problems.