Multidimensional Uncertainty Quantification via Optimal Transport

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Kotelevskii, Nikita, Goloburda, Maiya, Kondratyev, Vladimir, Fishkov, Alexander, Guizani, Mohsen, Moulines, Eric, Panov, Maxim
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866909809347919872
author Kotelevskii, Nikita
Goloburda, Maiya
Kondratyev, Vladimir
Fishkov, Alexander
Guizani, Mohsen
Moulines, Eric
Panov, Maxim
author_facet Kotelevskii, Nikita
Goloburda, Maiya
Kondratyev, Vladimir
Fishkov, Alexander
Guizani, Mohsen
Moulines, Eric
Panov, Maxim
contents Most uncertainty quantification (UQ) approaches provide a single scalar value as a measure of model reliability. However, different uncertainty measures could provide complementary information on the prediction confidence. Even measures targeting the same type of uncertainty (e.g., ensemble-based and density-based measures of epistemic uncertainty) may capture different failure modes. We take a multidimensional view on UQ by stacking complementary UQ measures into a vector. Such vectors are assigned with Monge-Kantorovich ranks produced by an optimal-transport-based ordering method. The prediction is then deemed more uncertain than the other if it has a higher rank. The resulting VecUQ-OT algorithm uses entropy-regularized optimal transport. The transport map is learned on vectors of scores from in-distribution data and, by design, applies to unseen inputs, including out-of-distribution cases, without retraining. Our framework supports flexible non-additive uncertainty fusion (including aleatoric and epistemic components). It yields a robust ordering for downstream tasks such as selective prediction, misclassification detection, out-of-distribution detection, and selective generation. Across synthetic, image, and text data, VecUQ-OT shows high efficiency even when individual measures fail. The code for the method is available at: https://github.com/stat-ml/multidimensional_uncertainty.
format Preprint
id arxiv_https___arxiv_org_abs_2509_22380
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Multidimensional Uncertainty Quantification via Optimal Transport
Kotelevskii, Nikita
Goloburda, Maiya
Kondratyev, Vladimir
Fishkov, Alexander
Guizani, Mohsen
Moulines, Eric
Panov, Maxim
Machine Learning
Most uncertainty quantification (UQ) approaches provide a single scalar value as a measure of model reliability. However, different uncertainty measures could provide complementary information on the prediction confidence. Even measures targeting the same type of uncertainty (e.g., ensemble-based and density-based measures of epistemic uncertainty) may capture different failure modes. We take a multidimensional view on UQ by stacking complementary UQ measures into a vector. Such vectors are assigned with Monge-Kantorovich ranks produced by an optimal-transport-based ordering method. The prediction is then deemed more uncertain than the other if it has a higher rank. The resulting VecUQ-OT algorithm uses entropy-regularized optimal transport. The transport map is learned on vectors of scores from in-distribution data and, by design, applies to unseen inputs, including out-of-distribution cases, without retraining. Our framework supports flexible non-additive uncertainty fusion (including aleatoric and epistemic components). It yields a robust ordering for downstream tasks such as selective prediction, misclassification detection, out-of-distribution detection, and selective generation. Across synthetic, image, and text data, VecUQ-OT shows high efficiency even when individual measures fail. The code for the method is available at: https://github.com/stat-ml/multidimensional_uncertainty.
title Multidimensional Uncertainty Quantification via Optimal Transport
topic Machine Learning
url https://arxiv.org/abs/2509.22380