Low-Budget Active Learning through Entropic Optimal Transport
Abstract
We consider low-budget active learning, which consists of selecting a limited number of points, the coreset, such that a model can be trained to high accuracy on the selection only. This problem is particularly relevant in contexts where labeling requires costly expert intervention, as in medical applications. We leverage features extracted from a pretrained self-supervised model to represent the data, and perform coreset selection directly in this feature space. In this paper, we use entropic optimal transport, specifically the Sinkhorn divergence, as the coreset selection criterion, which first allows us to get dimension-free sample complexity results, and second admits computationally efficient gradient evaluations. This opens the way to using gradient-based algorithms to rapidly compute solution candidates, further improved by a swap-based local search, with guarantees on the solution quality. Experiments on image benchmarks and medical datasets show that our method outperforms state-of-the-art heuristics in low-budget settings.
Related
- First-order Γ-expansion for entropic optimal transport with non-degenerate costs
- On the Linear Convergence of SISTA for Entropic Martingale Optimal Transport
- Projected McKean--Vlasov Dynamics for Entropic Weak Optimal Transport
- Statistical Rates for Entropic Optimal Transport in the Discrete to SubGaussian Regime
- Sequential Entropic Optimal Transport for Bayesian Inverse Problems via Multilevel Monte Carlo
BibTeX
@misc{hajal2026lowbudget,
title = {{Low-Budget Active Learning through Entropic Optimal Transport}},
author = {Rim Hajal and Mathieu Besançon and Jérôme Malick},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.01199},
eprint = {2610.01199},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.01199},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Low-Budget Active Learning through Entropic Optimal Transport
AU - Rim Hajal
AU - Mathieu Besançon
AU - Jérôme Malick
PY - 2026
DA - 2026/10/01
PB - arXiv
DO - 10.48550/arxiv.2610.01199
UR - https://arxiv.org/abs/2610.01199
AB - We consider low-budget active learning, which consists of selecting a limited number of points, the coreset, such that a model can be trained to high accuracy on the selection only. This problem is particularly relevant in contexts where labeling requires costly expert intervention, as in medical applications. We leverage features extracted from a pretrained self-supervised model to represent the data, and perform coreset selection directly in this feature space. In this paper, we use entropic optimal transport, specifically the Sinkhorn divergence, as the coreset selection criterion, which first allows us to get dimension-free sample complexity results, and second admits computationally efficient gradient evaluations. This opens the way to using gradient-based algorithms to rapidly compute solution candidates, further improved by a swap-based local search, with guarantees on the solution quality. Experiments on image benchmarks and medical datasets show that our method outperforms state-of-the-art heuristics in low-budget settings.
ER -