A Local Mixup to Prevent Manifold Intrusion - Archive ouverte HAL Access content directly
Conference Papers Year :

A Local Mixup to Prevent Manifold Intrusion

(1, 2) , (1, 3) , (1, 2)
1
2
3

Abstract

Deployed in the context of supervised learning, Mixup is a data-dependent regularization technique that consists in linearly interpolating input samples and associated outputs. It has been shown to improve accuracy when used to train on standard machine learning datasets. However, authors have pointed out that Mixup can produce out-of-distribution virtual samples and even contradictions in the augmented training set, potentially resulting in adversarial effects. In this paper, we introduce Local Mixup in which distant input samples are weighted down when computing the loss. In constrained settings we demonstrate that Local Mixup can create a trade-off between bias and variance, with the extreme cases reducing to vanilla training and classical Mixup. Using standardized computer vision benchmarks, we also show that Local Mixup can improve accuracy.
Fichier principal
Vignette du fichier
20220214042103_872102_1084.pdf (256.35 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03683702 , version 1 (31-05-2022)

Identifiers

  • HAL Id : hal-03683702 , version 1

Cite

Raphael Baena, Lucas Drumetz, Vincent Gripon. A Local Mixup to Prevent Manifold Intrusion. EUSIPCO 2022, Aug 2022, Belgrade, Serbia. ⟨hal-03683702⟩
35 View
36 Download

Share

Gmail Facebook Twitter LinkedIn More