Tailoring Mixup to Data using Kernel Warping functions - Equipe Signal, Statistique et Apprentissage Access content directly
Preprints, Working Papers, ... (Preprint) Year : 2023

Tailoring Mixup to Data using Kernel Warping functions

Abstract

Data augmentation is an essential building block for learning efficient deep learning models. Among all augmentation techniques proposed so far, linear interpolation of training data points, also called mixup, has found to be effective for a large panel of applications. While the majority of works have focused on selecting the right points to mix, or applying complex non-linear interpolation, we are interested in mixing similar points more frequently and strongly than less similar ones. To this end, we propose to dynamically change the underlying distribution of interpolation coefficients through warping functions, depending on the similarity between data points to combine. We define an efficient and flexible framework to do so without losing in diversity. We provide extensive experiments for classification and regression tasks, showing that our proposed method improves both performance and calibration of models.
Fichier principal
Vignette du fichier
main.pdf (3.38 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-04552478 , version 1 (19-04-2024)

Identifiers

Cite

Quentin Bouniot, Pavlo Mozharovskyi, Florence d'Alché-Buc. Tailoring Mixup to Data using Kernel Warping functions. 2024. ⟨hal-04552478⟩
108 View
13 Download

Altmetric

Share

Gmail Mastodon Facebook X LinkedIn More