We’re getting more and more used to streaming services recommending movies to see, or music to listen to, based on what we’ve consumed before. But the suggestions must be relevant to us, otherwise they are redundant.
Explainable machine learning is a research branch that aims at explaining the suggestions that machine learning delivers, whether it’s a medical recommendation or a suggestion to listen to a specific song. Researchers are thus trying to unbox the “black box” of machine learning.
Maria Movin, Guilherme Dinis Chaliane and Jonathan Piller are working with questions related to explainable machine learning from two angles. They share their time between business and academia, being employed by Spotify and at the same time being industrial doctoral students at the Department for Computer and Systems Sciences (DSV), Stockholm University.
Professor Panagiotis Papapetrou is their scientific supervisor at DSV. He is very positive to this type of collaboration with the private sector.
“It’s great that Spotify has chosen to collaborate with us at DSV. The industrial doctoral students bring interesting cases and data, and we shape research questions together. The methods that they develop can then be used in different settings. Important knowledge is generated for both parties”, says Panagiotis Papapetrou.
Contact panagiotis@dsv.su.se
A longer version of this article is available in Swedish
What is an industrial doctoral student?
A “normal” PhD student finishes their doctoral thesis in four years. The industrial doctoral students work with their theses part-time for up to eight years.
Spotify finances the PhD studies of Maria Movin, Guilherme Dinis Chaliane and Jonathan Piller. The Department for Computer and Systems Sciences at Stockholm University is responsible for their scientific education.
The industrial doctoral students have both academic and industrial supervisors.
At DSV, the students belong to the Data Science Research Group.