In this paper, we investigate if Deep Metric Learning models are prone to background bias and test a method to alleviate such bias.
MehrNews - Data Science Chair
Our paper "Towards Explainable Occupational Fraud Detection" has been accepted at MIDAS 2022
22.07.2022
In our paper, we investigate the performance and interpretability of machine learning approaches when detecting occupational fraud in company data, finding models that give both strong performance and comprehensible decisions.
Mehr
In our paper we use the Method of Lines together with Neural ODEs to learn spatio-temporal data arising from dynamical systems. We show that the method performs consistently well on a wide range of applications.
MehrWe've created an approach for constructing synthetic ERP system data with different cases of occupational fraud to help researchers develop automated detection approaches.
MehrOur paper "Semi-unsupervised Learning forTime Series Classification" got accepted @ MILETS (KDD22)
30.06.2022Our workshop paper "Semi-unsupervised Learning forTime Series Classification" got accepted @ MILETS (8th SIGKDD International Workshop on Mining and Learning from Time Series -- Deep Forecasting: Models, Interpretability, and Applications) at KDD22.
MehrWe are hiring!
29.06.2022We currently have two open positions for PhD students! Check here for more information on what you can do at our chair and how to apply.
MehrThe "NLP colloquium", a series of online research talks on methodological developments and interesting applications in/of Natural Language Processing will start this week. The colloquium is organized by the Chair for NLP (Prof. Glavaš), in collaboration with the Chairs for Data Science (Prof. Hotho) and Computational Philology (Prof. Jannidis).
Mehr
Our recent ICCV paper "Do Different Deep Metric Learning Losses Lead to Similar Learned Features?" is currently trending on the AI research discovery platform deepai.org.
Mehr
Das #Verkehrsministerium unterstützt das #P-BIM Projekt zur KI-gestützten Brückenüberwachung. Projektstart ist der 01.09.2022. Mit dem Projekt sollen Brückenzustände auf smarte Art und Weise ermittelt und Schäden frühzeitig erkannt werden.
MehrDetection of the pressure wave of the Hunga Tonga volcanic eruption with the we4bee hives
21.01.2022In our project we4bee, we detected two main air pressure waves in Germany from the Tonga eruption. One from north to south, the second one a few hours later from south to north. We visualized pressure differences for our top bar hive for both waves and interpolated isobares from all locations in Germany.
Mehr