Luka Biedebach
Luka Biedebach first joined the Sleep Revolution in 2021 to write her master thesis on identifying paediatric mouth breathing during sleep. During this time, she created an unsupervised reconstruction-based anomaly detection model that classifies breathing sequences. She then graduated from the University of Mannheim with a Master Degree in Data Science in 2022. Exploring all the possibilities of applying machine learning in sleep research raised her interest to dive deeper into this field.
In 2022 she came back to Reykjavik University as Ph.D. Candidate within the Sleep Revolution. Her research focus is on unsupervised machine learning in sleep research. She currently researches on applying unsupervised learning on sleep EEGs and sleep quality measurements from wearable sensors.
Research Interests
- Unsupervised Learning
- Multivariate Time Series Classification
- Autoencoders
- Generative Models
Teaching
- SLEEP (Spring Semester 2023)
- Digital Health (Fall Semester 2023)
Publications
Biedebach, L., Óskarsdóttir, M., Arnardottir, E. S., & Islind, A. S. (2023). Two Sides of the Same Pillow: Unfolding the Relationship between Objective and Subjective Sleep Quality with Unsupervised Learning. 44th International Conference on Information Systems.
Biedebach, L., Óskarsdóttir, M., Arnardóttir, E.S., Sigurdardóttir, S., Clausen, M. V., Sigurdardóttir, S. Þ., Serwatko, M. & Islind, A.S. (2023) Anomaly detection in sleep: detecting mouth breathing in children. Data Mining and Knowledge Discovery.
Biedebach, L., Rusanen, M., Þórðarson, B., Arnadóttir, E. S., Óskarsdóttir, M., Nikkonen, S., … & Islind, A. S. (2023).,Towards a Deeper Understanding of Sleep Stages through their Representation in the Latent Space of Variational Autoencoders. 56th Hawaii International Conference on System Sciences, 3111-3121.