Communication Dans Un Congrès Année : 2025

One for all and all for one: Efficient Computation of Partial Wasserstein Distances on the Line

Résumé

Partial Wasserstein helps overcoming some of the limitations of Optimal Transport when the distributions at stake differ in mass, contain noise or outliers or exhibit mass mismatches across distribution modes. We introduce PAWL, a novel algorithm designed to efficiently compute exact PArtial Wasserstein distances on the Line. PAWL not only solves the partial transportation problem for a specified amount of mass to be transported, but for all admissible ones. This flexibility is valuable for machine learning tasks where the level of noise is uncertain and needs to be determined through, e.g., cross-validation. By achieving O (n log n) time complexity for the partial 1-Wasserstein problem on the line, it enables practical applications with large scale datasets. Additionally, we introduce a novel slicing strategy tailored to Partial Wasserstein, which does not permit transporting mass between outliers or noisy data points. We demonstrate the advantages of PAWL in terms of computational efficiency and performance in downstream tasks, outperforming existing (sliced) Partial Optimal Transport techniques.
Fichier principal
Vignette du fichier
Reg_path_sliced_partial_OT__rebutal_ICLR_.pdf (1) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04928469 , version 1 (04-02-2025)
hal-04928469 , version 2 (10-02-2025)

Licence

Identifiants

  • HAL Id : hal-04928469 , version 2

Citer

Romain Tavenard, Laetitia Chapel. One for all and all for one: Efficient Computation of Partial Wasserstein Distances on the Line. International Conference on Learning Representations, Apr 2025, Singapore, Singapore. ⟨hal-04928469v2⟩
0 Consultations
0 Téléchargements

Partager

More