Contextual movement models based on normalizing flows

Publikation: Beiträge in ZeitschriftenZeitschriftenaufsätzeForschungbegutachtet

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.
OriginalspracheEnglisch
ZeitschriftAStA Advances in Statistical Analysis
Jahrgang107
Ausgabenummer1-2
Seiten (von - bis)51-72
Anzahl der Seiten22
ISSN1863-8171
DOIs
PublikationsstatusErschienen - 03.2023

DOI

Zuletzt angesehen

Forschende

  1. Neele Puhlmann

Publikationen

  1. Web-Based Drills in Maths Using a Computer Algebra System
  2. On the distinctiveness of tags in collaborative tagging systems
  3. A Geometric Approach by Using Switching and Flatness Based Control in Electromechanical Actuators for Linear Motion
  4. Deconstructing the Theoretical Language of Process Research
  5. Computational modeling of amorphous polymers
  6. An isomorphism between polynomial eigenfunctions of the transfer operator and the Eichler cohomology for modular groups
  7. Efficient co-regularised least squares regression
  8. Defining the notion of mining, extraction and collection
  9. Influence of Equal-Channel Angular Pressing on the Microstructure and Texture of Mg-Zn-Y-Zr-RE Alloy Sheets
  10. Tree diversity increases productivity through enhancing structural complexity across mycorrhizal types
  11. Multiphase-field modeling of temperature-driven intermetallic compound evolution in an Al-Mg system for application to solid-state joining processes
  12. Assessing Effects Through Semi-Field and Field Toxicity Testing
  13. Controlling a Bank Model Economy by Using an Adaptive Model Predictive Control with Help of an Extended Kalman Filter
  14. Introduction: The representative turn in EU studies
  15. Differences in the sophistication of Value-based Management
  16. Order parameters and energies of analytic and singular vortex lines in rotating3He-A
  17. Lost-customers approximation of semi-open queueing networks with backordering
  18. In situ synchrotron radiation diffraction investigation of the compression behaviour at 350 °C of ZK40 alloys with addition of CaO and Y
  19. Mechanical behavior, microstructural evolution and texture analysis of AA2024-T351 processed by multi-layer friction surfacing with high build rates
  20. Speed of processing and stimulus complexity in low-frequency and high-frequency channels
  21. Implicit and explicit horizons
  22. Diffusion patterns in small vs. large capital markets-the case of value-based management
  23. Supporting Visual and Verbal Learning Preferences in a Second-Language Multimedia Learning Environment
  24. (Re)productivity
  25. Comparison of Trajectory Estimation Methods Based on LIDAR and Monocular Camera in a Simulated Environment