aiqbench analysis
This paper gives statistical guarantees for training classifiers and regressors on synthetic functional data from a class-conditional diffusion generator, using Wasserstein-1 distance because total variation is uninformative for curves. The results bound the risk of synthetic-augmented training against real-only training and derive an optimal synthetic sample size, while the experiments on temperature, hyperspectral and gait curves are modest in scope. For ML practitioners working with small datasets, the useful takeaway is the conditions under which synthetic data helps and how to allocate it under class imbalance.
Synthetic Data Augmentation for Functional Data: Wasserstein Risk Bounds and Diffusion-Based Conditional Generation
arXiv:2610.34120v1 Announce Type: cross Abstract: Functional data are costly to collect; samples of tens to hundreds of curves are common. Synthetic data augmentation can help, but its theory is developed for finite-dimensional data and relies on total variation, which is uninformative for functions. We study augmentation with a class-conditional functional diffusion generator and state guarantees in the Wasserstein-1 distance. For stable downstream estimators, we bound the risk of training on synthetic curves, from above and below, by the risk of training on as many real curves plus a generation penalty, and derive the best number of synthetic curves and a condition under which synthetic data outperform real data. For allocating synthetic curves under class imbalance, we bound the excess risk of training on real and synthetic curves together, without sample splitting. In applications to temperature, hyperspectral and gait curves, augmentation improved classification, function-to-scalar regression and function-to-function regression, and directing synthetic curves to the minority class gave the largest gain under class imbalance.
Abstract:Functional data are costly to collect; samples of tens to hundreds of curves are common. Synthetic data augmentation can help, but its theory is developed for finite-dimensional data and relies on total variation, which is uninformative for functions. We study augmentation with a class-conditional functional diffusion generator and state guarantees in the Wasserstein-1 distance. For stable downstream estimators, we bound the risk of training on synthetic curves, from above and below, by the risk of training on as many real curves plus a generation penalty, and derive the best number of synthetic curves and a condition under which synthetic data outperform real data. For allocating synthetic curves under class imbalance, we bound the excess risk of training on real and synthetic curves together, without sample splitting. In applications to temperature, hyperspectral and gait curves, augmentation improved classification, function-to-scalar regression and function-to-function regression, and directing synthetic curves to the minority class gave the largest gain under class imbalance.
Methodology (stat.ME); Machine Learning (cs.LG); Machine Learning (stat.ML)
arXiv:2610.34120 [stat.ME]
(or arXiv:2610.34120v1 [stat.ME] for this version)
https://doi.org/10.48550/arXiv.2610.34120 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 256 words