Machine Learning for Signal Processing

Basic methods for statistical pattern recognition/machine learning. Deep neural networks, support vector machines, mixture models, hidden Markov models. Design, training and evaluation of machine learning models. Extraction of feature vectors with applications to speech technology, medical signal processing and multimedia signal processing.

Knowledge The candidate has - good understanding of the theoretical principles and practical aspects of using statistical pattern recognition/machine learning - good understanding of best practice with regards to the use of training, validation and test data - broad knowledge on the properties of speech, medical and multimedia signals - broad knowledge on feature extraction for wide variety of signals Skill: The candidate can - use and/or design software for use in train and evaluate models based on machine learning methods - evaluate the performance of machine learning systems General competence: The candidate can - the insights in the interplay between basis technology and development of machine learning systems - conduct teamwork and documentation

Lectures, mandatory computer exercises. Note: in the Autumn term 2026, the main teacher of the course will be on sabbatical. Some of the lectures may only be available as recorder videos.