LIMMT: Less Is More for Motion Tracking

Yu Guan, Zekun Qi, Chenghuai Lin, Xuchuan Chen, Wenyao Zhang, Jilong Wang, Xinqiang Yu, He Wang, Li Yi
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:37504-37517, 2026.

Abstract

We argue that high-quality motion data can steer tracking policies toward better optimization trajectories early in training. In this work, we introduce LIMMT (Less Is More for Motion Tracking). To our knowledge, this is the first data-centric study for physics-based humanoid motion tracking. We go beyond simply removing erroneous clips. We define motion data quality through three dimensions: physics feasibility, diversity, and complexity. We show that training with under 3% of AMASS yields better tracking performance than training with the full dataset. Extensive experiments and analyses validate the effectiveness of our framework.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-guan26f, title = {{LIMMT}: Less Is More for Motion Tracking}, author = {Guan, Yu and Qi, Zekun and Lin, Chenghuai and Chen, Xuchuan and Zhang, Wenyao and Wang, Jilong and Yu, Xinqiang and Wang, He and Yi, Li}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {37504--37517}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/guan26f/guan26f.pdf}, url = {https://proceedings.mlr.press/v306/guan26f.html}, abstract = {We argue that high-quality motion data can steer tracking policies toward better optimization trajectories early in training. In this work, we introduce LIMMT (Less Is More for Motion Tracking). To our knowledge, this is the first data-centric study for physics-based humanoid motion tracking. We go beyond simply removing erroneous clips. We define motion data quality through three dimensions: physics feasibility, diversity, and complexity. We show that training with under 3% of AMASS yields better tracking performance than training with the full dataset. Extensive experiments and analyses validate the effectiveness of our framework.} }
Endnote
%0 Conference Paper %T LIMMT: Less Is More for Motion Tracking %A Yu Guan %A Zekun Qi %A Chenghuai Lin %A Xuchuan Chen %A Wenyao Zhang %A Jilong Wang %A Xinqiang Yu %A He Wang %A Li Yi %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-guan26f %I PMLR %P 37504--37517 %U https://proceedings.mlr.press/v306/guan26f.html %V 306 %X We argue that high-quality motion data can steer tracking policies toward better optimization trajectories early in training. In this work, we introduce LIMMT (Less Is More for Motion Tracking). To our knowledge, this is the first data-centric study for physics-based humanoid motion tracking. We go beyond simply removing erroneous clips. We define motion data quality through three dimensions: physics feasibility, diversity, and complexity. We show that training with under 3% of AMASS yields better tracking performance than training with the full dataset. Extensive experiments and analyses validate the effectiveness of our framework.
APA
Guan, Y., Qi, Z., Lin, C., Chen, X., Zhang, W., Wang, J., Yu, X., Wang, H. & Yi, L.. (2026). LIMMT: Less Is More for Motion Tracking. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:37504-37517 Available from https://proceedings.mlr.press/v306/guan26f.html.

Related Material