New Publication: “Active Learning of Ordinal Embeddings: A User Study on Football Data”

We are happy to share the news that our user study on Active Learning of Ordinal Embeddings by Löffler et al. is now published in the Transactions on Machine Learning Research (TMLR), see OpenReview. This work improves upon our prior work on information retrieval in a football trajectory dataset (see doi/10.1145/3465057). In our new paper, we use deep metric learning and an entropy-based active learning method, and analyze the effectiveness of sampling heuristics through a user study. In short, we can not only retrieve similar football scenes very quickly, but can now also quickly learn a “similarity metric”, that is, the user’s own perception of what “similar” means.

We summarize the paper in a short video presentation, and share the code to run a browser-based user study (see Github). We want to thank our friends at the Fraunhofer IIS’s ADA Lovelace Center and the Georgia Tech’s SIPLab for the great collaboration. For more information, see the project website.

We would like to acknowledge support for this project from the IFI programme of the German Academic Exchange Service (DAAD), the Bavarian Ministry of Economic Affairs, Infrastructure, Energy and Technology as part of the Bavarian project Leistungszentrum Elektroniksysteme (LZE) and the Center for Analytics-Data-Applications (ADA-Center) within the framework of “BAYERN DIGITAL II”. We thank the study participants for their valuable contributions.