AUTHOR=Oppel Heiko , Spilz Andreas , Munz Michael 

TITLE=Optimizing training of time series diffusion models via similarity score functions: application to cyclic and acyclic motion with IMU data

JOURNAL=Frontiers in Artificial Intelligence

VOLUME=Volume 8 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/artificial-intelligence/articles/10.3389/frai.2025.1640948

DOI=10.3389/frai.2025.1640948

ISSN=2624-8212

ABSTRACT=IntroductionDenoising diffusion probabilistic models have shown the capability to generate synthetic sensor signals. These models rely on a loss function that measures the difference between the noise added during the forward process and the noise predicted by the diffusion model, thereby enabling realistic data generation. However, the stochastic nature of the process and the loss function complicate the estimation of data quality.MethodsTo address this issue, we evaluated multiple similarity metrics and adapted an existing metric to monitor both the training and data synthesis processes. The adapted metric was further fine-tuned on the input data to align with the requirements of a downstream classification task.ResultsBy incorporating the adapted metric, we significantly reduced the number of training epochs required without observing performance degradation in the classification task.DiscussionOur findings demonstrate that optimizing the training process using similarity metrics not only conserves computational resources but also shortens the training time for generative models, making them more efficient and practical for real-world applications.