Semi-Supervised Generative Models for Multi-Agent Trajectories

Research output: Contributions to collected editions/worksArticle in conference proceedingsResearchpeer-review

Authors

Analyzing the spatiotemporal behavior of multiple agents is of great interest to many communities. Existing probabilistic models in this realm are formalized either in an unsupervised framework, where the latent space is described by discrete or continuous variables, or in a supervised framework, where weakly preserved labels add explicit information to continuous latent representations. To overcome inherent limitations, we propose a novel objective function for processing multi-agent trajectories based on semi-supervised variational autoencoders, where equivariance and interaction of agents are captured via customized graph networks. The resulting architecture disentangles discrete and continuous latent effects and provides a natural solution for injecting expensive domain knowledge into interactive sequential systems. Empirically, our model not only outperforms various state-of-the-art baselines in trajectory forecasting, but also learns to effectively leverage unsupervised multi-agent sequences for classification tasks on interactive real-world sports datasets.

Original languageEnglish
Title of host publicationAdvances in Neural Information Processing Systems 35 : 36th Conference on Neural Information Processing Systems (NeurIPS 2022)
EditorsS. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, A. Oh
Number of pages15
Volume48
Place of PublicationRed Hook
PublisherCurran Associates
Publication date2022
Pages37267-37281
ISBN (print)978-1-7138-7108-8
ISBN (electronic)978-1-7138-7312-9
Publication statusPublished - 2022
Event36th Conference on Neural Information Processing Systems - NeurIPS 2022 - New Orleans Convention Center, New Orleans, United States
Duration: 28.11.202209.12.2022
Conference number: 36
https://nips.cc/Conferences/2022

Bibliographical note

Publisher Copyright:
© 2022 Neural information processing systems foundation. All rights reserved.

Recently viewed

Activities

  1. Towards a Techno-Ecology of Participation - 2017
  2. Teams are changing! Going into the wild to expand theory on dynamics in modern teamwork settings
  3. Where To Start? Exploring 1-Year-Students’ Preconceptions of Sustainable Development
  4. Knowledge Spaces
  5. Lena Meyer-Bergner’s conception of modernism between graphics and weaving, between folk art and technology
  6. Monomers release from composite materials after halogen and LED curing.
  7. Coherent behavior in geophysical flows
  8. Dynamics Days Europe 2018
  9. Intersecting Practices and Experiences of Educational Exclusions in Germany and Turkey – Implications for Transnational Professionalisation Processes
  10. Institute for Advanced Study Princeton
  11. UV photodegradation of trimipramine under different environmental variables and chemical nature of aqueous solution - biodegradation and LC-MSn characterization of the formed transformation products
  12. Rudimentary theory of entrepreneurship
  13. Grünes Design allein reicht nicht mehr
  14. Performing ventures in, trough and with regional start-up ecosystems: Emering insights on slowness in accelerators and the pedagogy of pitching
  15. Rational Design of Molecules by Life Cycle Engineering.
  16. Measurement of Perceived Mental Strain and Physical Exertion Using the Category Partitioning Procedure
  17. Do we need a new paradigm for mastering existing and future challenges of the urban water cycle
  18. Ludic Overload - Ludic Overkill: Gamification in the age of Media Overload
  19. Sustainability Reporting and the Need for Capacity Development
  20. Keynote speech entitled: Kalman Filters as Virtual Sensors and their Applications
  21. PRIORITIZATION OF VETERINARY ANTIBIOTICS FOR ENVIRONMENTAL ANALYSIS USING A SIMPLE SCREENING APPROACH
  22. “Relying on Spontaneity”
  23. Small Formation: Lines and Alternating Knots as Logistic Inversions

Publications

  1. Binary Random Nets II
  2. A Process Perspective on Organizational Failure
  3. Developing robust field survey protocols in landscape ecology
  4. MICSIM: Concept, Developments, and Applications of a PC Microsimulation Model for Research and Teaching
  5. CHANGING RECREATIONAL ACTIVITIES FOR REDUCING INSOMNIA SEVERITY? RESULTS FROM A SERIAL MEDIATION ANALYSIS ON THE IMPACT OF RECREATIONAL BEHAVIOR AS A MECHANISM OF CHANGE IN DIGITAL INTERVENTIONS FOR INSOMNIA
  6. Design of an Information-Based Distributed Production Planning System
  7. Continental mapping of forest ecosystem functions reveals a high but unrealised potential for forest multifunctionality.
  8. Wozu in Tönen denken?
  9. Web-Based Drills in Maths Using a Computer Algebra System
  10. Pushing the Envelope: Creating Public Value in the Labor Market
  11. Evaluating the effectiveness of retention forestry to enhance biodiversity in production forests of Central Europe using an interdisciplinary, multi-scale approach
  12. Tree diversity alters the structure of a tri-trophic network in a biodiversity experiment
  13. Big Data - Characterizing an Emerging Research Field using Topic Models
  14. Assessing mire-specific biodiversity with an indicator based approach
  15. Introduction to the Special Issue Section
  16. Machine Learning and Data Mining for Sports Analytics
  17. Editorial: Machine Learning and Data Mining in Materials Science
  18. (How) Can didactic research find its way into the classroom? Results from a questionnaire survey on the lesson preparation and continuing professional development of German teachers
  19. Laser Scanning Point Cloud Improvement by Implementation of RANSAC for Pipeline Inspection Application
  20. Sustainable use of ecosystem services under multiple risks
  21. Collaborative business in supply chains - a system dynamics approach
  22. Using Reading Strategy Training to Foster Students´ Mathematical Modelling Competencies
  23. Dynamic efficiency and path dependencies in venture capital markets
  24. A review of mobile language learning applications
  25. Error handling in office work with computers
  26. Challenging the status quo of accelerator research: Concluding remarks