Skip to main content

Henrik Boström

Title of the project
Reliable Machine Learning 

Background and summary of fellowship:
In many application areas, it is not sufficient to present the output of machine learning models to the users without providing any information on what leads to the specific predictions or recommendations and how (un)certain they are. The strongest machine learning models are however often essentially black-boxes. In order to enable trust in such models, techniques for explaining the predictions in the form of interpretable approximations are currently being investigated. Another cornerstone for enabling trust is that the uncertainty of the output of the machine learning models is properly quantified, e.g., that the output prediction intervals or probability distributions are well-calibrated. 

Motivated by collaborations with the Swedish National Financial Management Authority on gross domestic product (GDP) forecasting, with Karolinska Institute/University hospital on sepsis prediction, and Scania on predictive maintenance, techniques for quantifying uncertainty and explaining predictions of multivariate time-series forecasting models will be developed and evaluated. In addition to scientific papers, the output of the project will be Python packages to support reliable multi-variate time-series forecasting, enabling predictions of state-of-the-art machine learning models to be complemented with explanations and uncertainty quantification.


Picture of Henrik Boström

Henrik Boström

Professor, Division of Software and computer Systems at KTH, Working group Learn at Digital Futures, Digital Futures fellow

+46 8 790 43 06