Contextual movement models based on normalizing flows

Research output: Journal contributionsJournal articlesResearchpeer-review

Authors

Movement models predict positions of players (or objects in general) over time and are thus key to analyzing spatiotemporal data as it is often used in sports analytics. Existing movement models are either designed from physical principles or are entirely data-driven. However, the former suffers from oversimplifications to achieve feasible and interpretable models, while the latter relies on computationally costly, from a current point of view, nonparametric density estimations and require maintaining multiple estimators, each responsible for different types of movements (e.g., such as different velocities). In this paper, we propose a unified contextual probabilistic movement model based on normalizing flows. Our approach learns the desired densities by directly optimizing the likelihood and maintains only a single contextual model that can be conditioned on auxiliary variables. Training is simultaneously performed on all observed types of movements, resulting in an effective and efficient movement model. We empirically evaluate our approach on spatiotemporal data from professional soccer. Our findings show that our approach outperforms the state of the art while being orders of magnitude more efficient with respect to computation time and memory requirements.
Original languageEnglish
JournalAStA Advances in Statistical Analysis
Volume107
Issue number1-2
Pages (from-to)51-72
Number of pages22
ISSN1863-8171
DOIs
Publication statusPublished - 03.2023

Bibliographical note

Publisher Copyright:
© 2021, The Author(s).

    Research areas

  • Density estimation, Movement models, Normalizing flows, Soccer data, Spatiotemporal data, Sports analytics
  • Business informatics

Recently viewed

Activities

  1. Expertise in law: 'from above' and 'from below'
  2. Employer Longevity Readiness Index Workshop: Session 2: How do you build a longevity readiness Index?
  3. Travelling Codes
  4. Research Workshop “Innovation and Value Creation” - 2009
  5. How does tree sapling diversity influence browsing intensity by deer across spatial scales?
  6. On Borders, Boundaries, Clouds, and Globalization. And on China.
  7. ‘Thinking the Problematic‘
  8. Digitalization and Organizational Learning: Use the Double-Loop
  9. Academy of Management Annual Meeting 2023
  10. Harvard Universität
  11. Mathematical and Computational Applications (Fachzeitschrift)
  12. Project Workshop on "Worker Flows, Match Quality, and Productivity" - 2019
  13. Curating Diversity in Global Performance Art
  14. Evaluation of tension-compression asymmetry in nanocrystalline PdAu using a Drucker-Prager type constitutive model.
  15. Symposium "Art and its Frames - Continuity and Change" 2014
  16. Comfort and Intervention Behavior of Drivers in Highly Automated Vehicles with Headway Control
  17. International Conference of EAS and ISME - 2007
  18. Migrations of Knowledge - Migknow 2014
  19. The Predictive Power of Social Media Sentiment for Short-Term Stock Movements
  20. Co-creating transformative processes - a designerly approach
  21. Education for Sustainable Development – Experiences from Theory and Practice
  22. Requests in Nigerian and British English conversational interactions: A corpus-based approach.
  23. Research Workshop “Innovation & Value Creation"
  24. 3rd International Conference on Innovations in Bio-Inspired Computing and Applications: Program Committee Member - IBICA2012
  25. Prototypes: The Usefulf Ambiguity of the „Biological Computer" (Annual Meeting of the AMERICAN SOCIETY FOR CYBERNETICS)
  26. Video or Text Cases in Problem-Oriented or Direct Instructional Settings for Preservice Teachers?