High-Performance Self-Supervised Learning by Joint Training of Flow Matching

Kosuke Ukita, Tsuyoshi Okita
Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, PMLR 300:4492-4500, 2026.

Abstract

Diffusion models can learn rich representations during data generation, showing potential for Self-Supervised Learning (SSL), but they face a trade-off between generative quality and discriminative performance. Their iterative sampling also incurs substantial computational and energy costs, hindering industrial and edge AI applications. To address these issues, we propose the Flow Matching-based Sensor Foundation Model (SenFlow), which jointly trains a representation encoder and a conditional flow matching generator. This decoupled design achieves both high-fidelity generation and effective recognition. By using flow matching to learn a simpler velocity field, SenFlow accelerates and stabilizes training, improving its efficiency for representation learning. Experiments on wearable sensor data show SenFlow reduces training time by 50.4% compared to a diffusion-based approach. On downstream tasks, SenFlow surpassed the state-of-the-art SSL method on all five datasets while achieving up to a 51.0x inference speedup and maintaining high generative quality. The implementation code is available at \url{https://github.com/Okita-Laboratory/SenFlow.}

Cite this Paper

BibTeX
@InProceedings{pmlr-v300-ukita26a, title = { High-Performance Self-Supervised Learning by Joint Training of Flow Matching }, author = {Ukita, Kosuke and Okita, Tsuyoshi}, booktitle = {Proceedings of The 29th International Conference on Artificial Intelligence and Statistics}, pages = {4492--4500}, year = {2026}, editor = {Khan, Emtiyaz and Li, Yingzhen and Solin, Arno and Ramdas, Aaditya}, volume = {300}, series = {Proceedings of Machine Learning Research}, month = {02--05 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v300/main/assets/ukita26a/ukita26a.pdf}, url = {https://proceedings.mlr.press/v300/ukita26a.html}, abstract = { Diffusion models can learn rich representations during data generation, showing potential for Self-Supervised Learning (SSL), but they face a trade-off between generative quality and discriminative performance. Their iterative sampling also incurs substantial computational and energy costs, hindering industrial and edge AI applications. To address these issues, we propose the Flow Matching-based Sensor Foundation Model (SenFlow), which jointly trains a representation encoder and a conditional flow matching generator. This decoupled design achieves both high-fidelity generation and effective recognition. By using flow matching to learn a simpler velocity field, SenFlow accelerates and stabilizes training, improving its efficiency for representation learning. Experiments on wearable sensor data show SenFlow reduces training time by 50.4% compared to a diffusion-based approach. On downstream tasks, SenFlow surpassed the state-of-the-art SSL method on all five datasets while achieving up to a 51.0x inference speedup and maintaining high generative quality. The implementation code is available at \url{https://github.com/Okita-Laboratory/SenFlow.} } }
Endnote
%0 Conference Paper %T High-Performance Self-Supervised Learning by Joint Training of Flow Matching %A Kosuke Ukita %A Tsuyoshi Okita %B Proceedings of The 29th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2026 %E Emtiyaz Khan %E Yingzhen Li %E Arno Solin %E Aaditya Ramdas %F pmlr-v300-ukita26a %I PMLR %P 4492--4500 %U https://proceedings.mlr.press/v300/ukita26a.html %V 300 %X Diffusion models can learn rich representations during data generation, showing potential for Self-Supervised Learning (SSL), but they face a trade-off between generative quality and discriminative performance. Their iterative sampling also incurs substantial computational and energy costs, hindering industrial and edge AI applications. To address these issues, we propose the Flow Matching-based Sensor Foundation Model (SenFlow), which jointly trains a representation encoder and a conditional flow matching generator. This decoupled design achieves both high-fidelity generation and effective recognition. By using flow matching to learn a simpler velocity field, SenFlow accelerates and stabilizes training, improving its efficiency for representation learning. Experiments on wearable sensor data show SenFlow reduces training time by 50.4% compared to a diffusion-based approach. On downstream tasks, SenFlow surpassed the state-of-the-art SSL method on all five datasets while achieving up to a 51.0x inference speedup and maintaining high generative quality. The implementation code is available at \url{https://github.com/Okita-Laboratory/SenFlow.}
APA
Ukita, K. & Okita, T.. (2026). High-Performance Self-Supervised Learning by Joint Training of Flow Matching . Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 300:4492-4500 Available from https://proceedings.mlr.press/v300/ukita26a.html.

Related Material