TY - JOUR

T1 - Masked autoencoder for multiagent trajectories

AU - Rudolph, Yannick

AU - Brefeld, Ulf

N1 - Part of 1 collection: Special Issue on Machine Learning in Soccer

PY - 2025

Y1 - 2025

N2 - Automatically labeling trajectories of multiple agents is key to behavioral analyses but usually requires a large amount of manual annotations. This also applies to the domain of team sport analyses. In this paper, we specifically show how pretraining transformer models improves the classification performance on tracking data from professional soccer. For this purpose, we propose a novel self-supervised masked autoencoder for multiagent trajectories to effectively learn from only a few labeled sequences. Our approach builds upon a factorized transformer architecture for multiagent trajectory data and employs a masking scheme on the level of individual agent trajectories. As a result, our model allows for a reconstruction of masked trajectory segments while being permutation equivariant with respect to the agent trajectories. In addition to experiments on soccer, we demonstrate the usefulness of the proposed pretraining approach on multiagent pose data from entomology. In contrast to related work, our approach is conceptually much simpler, does not require handcrafted features and naturally allows for permutation invariance in downstream tasks.

AB - Automatically labeling trajectories of multiple agents is key to behavioral analyses but usually requires a large amount of manual annotations. This also applies to the domain of team sport analyses. In this paper, we specifically show how pretraining transformer models improves the classification performance on tracking data from professional soccer. For this purpose, we propose a novel self-supervised masked autoencoder for multiagent trajectories to effectively learn from only a few labeled sequences. Our approach builds upon a factorized transformer architecture for multiagent trajectory data and employs a masking scheme on the level of individual agent trajectories. As a result, our model allows for a reconstruction of masked trajectory segments while being permutation equivariant with respect to the agent trajectories. In addition to experiments on soccer, we demonstrate the usefulness of the proposed pretraining approach on multiagent pose data from entomology. In contrast to related work, our approach is conceptually much simpler, does not require handcrafted features and naturally allows for permutation invariance in downstream tasks.

KW - Business informatics

KW - Self-supervised learning

KW - Multiagent trajectories

KW - Masked autoencoder

KW - Transformer

KW - Tracking data

KW - Soccer

UR - https://link.springer.com/article/10.1007/s10994-024-06647-3

U2 - 10.1007/s10994-024-06647-3

DO - 10.1007/s10994-024-06647-3

M3 - Journal articles

VL - 114

JO - Machine Learning

JF - Machine Learning

SN - 0885-6125

IS - 2

M1 - 44

ER -