Principles and Theory for Machine Learning

Principles and Theory for Machine Learning

Overview

Description

• Foundations of supervised learning
• Optimization for ML
• Unsupervised learning
• Neural networks
• Deep learning
• Deep generative models
• Some ML weaknesses
• Interpretable-explainable AI
• Self-supervised learning and foundation models

See also Stud.IP entry

Qualification goals

After successfully completing this module, students should be able to

• understand and correctly apply basic concepts of machine learning,
• master elementary tools for analysing the performance of machine learning approaches,
• recognise the main limitations of machine learning methods,
• propose strategies to overcome such limitations.

Proof of Performance

1 examination: written exam, 90 minutes, or oral exam, 30 minutes, or take-home exam

1 academic achievement: 50% of the exercises must be passed

Literature

• Understanding Machine Learning, Shalev-Schwartz & Ben-David, 2014
• Learning Theory from First Principles, Bach, 2024
• Deep Learning, Goodfellow et al., 2016
• Mathematical Theory of Deep Learning, Petersen & Zech, 2024
• Mathematics for Machine Learning, Deisenroth et al., 2020
• Neural Networks and Deep Learning, Aggarwal, 2023 (2nd edition)
• Deep Learning Architectures, Calin, 2020