• We are very excited to welcome Prof. Josephine Thomas, who has been appointed as Professor of Machine Learning at our institute. Prof. Thomas conducts research in the field of graph neural networks, and we look forward to working with her in the future.

  • Teodor Stoev presented our work in progress paper "Text2RLab: No-Code Methodology for Robotic Programming and Interaction in Laboratory Tasks" at the ACM Intelligent User Interfaces (IUI'25) conference. The paper could be found here: Text2RLab: No-Code Methodology for Robotic Programming and Interaction in Laboratory Tasks.
     
  • The Institute of Data Science was represented at the university’s 2025 Partnertage alongside the Federal Ministry of the Interior, Building, and Digitalization. During an interactive poster and demo session, we showcased two of our latest research projects. Philipp Adämmer presented the app 'Tin Gustaf – Your Holy Land AI Pocket Museum', a joint project with the Gustaf Dalman Institute at the Faculty of Theology and the university's computer centre. The app features an AI-driven categorisation of the Gustaf Dalman Collection, which contains artefacts and photographs from the cultural landscape of Palestine around 1900. In addition, Florian Wagner demonstrated an enhanced version of our digital coffee maker, highlighting our continued interest in integrating sensor technology with AI. The event was a great success, with a well-organised structure that encouraged meaningful networking and discussion.

  • Kristina Yordanova will give the talk "On the Development of AI and the Danger of Gender Bias" within the lecture series “Data Feminism – Perspectives from Culture, Medicine & Computer Science”.

     
    Abstract: There are different types of data, such as textual, video, image or sensor data that are used for training artificial intelligence (AI) models. These data are almost always biased in a certain way. Some examples of bias are gender, ethnical or age-related bias. The bias in the data is then integrated in the AI system and it in turn acts and makes decisions according to the data (and bias) it was trained with. As AI, with its sub-field of machine learning (ML) gets more and more popular and AI applications become part of our everyday life, it is very important to mitigate the effect of bias. In this talk I first discuss how bias is introduced in the data and later on in AI systems. I will then illustrate this with some examples of AI systems and how they are affected by gender bias. Finally, I will discuss strategies for mitigating (gender) bias in AI systems.  
     

  • We are pleased to announce that our demo paper "Low-Cost Sensor-based Error Detection System for Coffee Preparation: A Demonstration." has been accepted for the PerCom conference! We will showcases a system that leverages Computational State Space Models (CSSM) to detect and correct user errors during a coffee cooking procedure. The system is designed to assist elderly individuals with the daily living task of coffee preparation, thereby enhancing their independence and quality of life. The key benefits of our architecture include its low-cost anonymous sensors, small activity recognition and error detection model, and no need for training data, making it a viable solution for widespread adoption.

  • Sumaiya suravee presented our paper titled, “Cross-Lingual Named Entity Recognition for Low-Resource Languages: A Hindi-Nepali Case Study Using Multilingual BERT Models” at the 4th Multilingual Representation Learning workshop of EMNLP2024. The paper is available online at ACL Anthology.

  • The Institute of Data Science proudly represented the University of Greifswald at the celebration event for the German Unity Day in Schwerin, showcasing comprehensible insights into the groups research. With a Demonstrator of a digital assistance system for daily routines, the possibilities of such systems in the everyday life of the people was presented. With the help of the fantastic, interdisciplinary team, the day was a complete success and an important step for the public perception of the university.
  • Teodor Stoev attended the Institute of Systems and Robotics at the University of Coimbra, Portugal (ISR-UC) and the team led by Prof. Dr. Cristiano Premebida between 23.11.24 and 27.11.24. This research visit was part of the Xpro Project, funded by the DAAD and led by Dr. -Ing. Martin Aleksandrov and Prof. Dr. -Ing. Kristina Yordanova. During his stay, Teodor worked on the improvement of a pipeline for human detection using the combination of RGB camera data, LiDAR sensors as well as temperature sensors (dataset available here). Furthermore, potential future collaborations were explored regarding the application of Bayesian methods for human detection using multimodal sensor data.
  • Last week at #InformatikFESTIVAL we had a very interesting hybrid workshop, with two keynote speakers on topics addressing data publishing and access for AI applications and understanding the EU AI act, two discussion sessions on data management and preservation for applied ML systems and from the AI perspective, numerous presentations of accepted papers and very engaging audience. All in all, it has been a very successful #ARDUOUSWorkshop instance, which has convinced us that we should continue to push forward the effort of building a community around the topic of data annotation and its influence on the whole data lifecycle. The recording of the workshop is currently available on the InformatikFESTIVAL platform and on the YouTube channel of GI.
  • Teodor Stoev delivered a talk on "Verwendung großer Sprachmodelle für Datenannotation" ("Using Large Language Models for Data Annotation") at the Big Data 3.0 — LLMs Networking Workshop organized by KI-MV.
  • Dipendra Yadav has been selected for the Innovators Connect – Tandem 2024 program, organized by the German Centre for Research and Innovation (DWIH) New Delhi, promoting collaboration between Indian and German innovators.
  • We were part of the NØRD conference, which took place on May 29-30 in Rostock. We presented different AI methods and technologies developed at the Institute of Data Science.
  • On June 27 - 28 Kristina Yordanova gave a keynote at the Health Longevity Innovation Days at the University of Zurich, where she talked about supporting healthy ageing trough AI methods for within-person analytics
  • Kristina Yordanova will present "Extracting Dynamic Semantic Structures from Heterogeneous Data" at Institute for Physics, University of Greifswald on 06.06.2024 at 16:00. The talk will cover AI methods combining symbolic modelling, probabilistic reasoning, and machine learning for dynamic system modelling and applications in behaviour monitoring, cognitive impairments, healthy ageing, and intelligent tutoring.
  • Clémence Réda, a Marie Skłodowska-Curie Postdoctoral Fellow at the University of Rostock and mentee of our professor Kristina Yordanova, visited our Institute on May 8, 2024, providing insights into her current research and engaging in discussions with our researchers on the latest advancements and challenges in data science.
  • The Department of Data Science is proud to announce that it will be organizing the 8th International Workshop on Annotation of useR Data for UbiquitOUs Systems (ARDUOUS 2024). The workshop will be held on September 24, 2024, at the INFORMATIK FESTIVAL in Wiesbaden, Germany. For more information about the workshop, please visit the official website.
  • We're delighted to announce the addition of Florian Wagner to our department. With a Master's in Computer Science, Florian will be working on conceptualizing and administering our new Sensorlab, with a special focus on enhancing digital infrastructure and connectivity with the university's computer center.
  • Our department participated in "Research Colloquium: Reliable Machine Learning" organized by the Institute for Visual & Analytic Computing at the University of Rostock; Sumaiya Suravee presented her research on "Ontology development and methods for graph learning from textual data."
  • Our paper titled, "Enhancing Kitchen Activity Recognition: A Benchmark Study of the Rostock KTA Dataset," is accepted in IEEE Access, vol. 12, pp. 14364-14384, 2024, doi: 10.1109/ACCESS.2024.3356352.
  • Kristina Yordanova held a talk at the Late Lunch Seminar in Greifswald University Medicine (UMG) on 8. November 2023.
  • Our paper "Sensor-based Locomotion Data Mining for Supporting the Diagnosis of Neurodegenerative Disorders: a Survey" has been accepted in ACM Computing Surveys (Impact Factor 14.324)! You can find the paper here: https://dl.acm.org/doi/10.1145/3603495
  • Dipendra Yadav will present his PhD concept on "Evaluating Dangerous Capabilities of Large Language Models: An Examination of Situational Awareness" at this year's KI doctoral consortium, which will take place in Berlin from 26 - 29 September.