Data Augmentation with Variational Autoencoders and Manifold Sampling

Clément Chadebec, Stéphanie Allassonnière

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

We propose a new efficient way to sample from a Variational Autoencoder in the challenging low sample size setting (A code is available at https://github.com/clementchadebec/Data_Augmentation_with_VAE-DALI ). This method reveals particularly well suited to perform data augmentation in such a low data regime and is validated across various standard and real-life data sets. In particular, this scheme allows to greatly improve classification results on the OASIS database where balanced accuracy jumps from 80.7% for a classifier trained with the raw data to 88.6% when trained only with the synthetic data generated by our method. Such results were also observed on 3 standard data sets and with other classifiers.

Original languageEnglish
Title of host publicationDeep Generative Models, and Data Augmentation, Labelling, and Imperfections - First Workshop, DGM4MICCAI 2021, and First Workshop, DALI 2021, Held in Conjunction with MICCAI 2021, Proceedings
EditorsSandy Engelhardt, Ilkay Oksuz, Dajiang Zhu, Yixuan Yuan, Anirban Mukhopadhyay, Nicholas Heller, Sharon Xiaolei Huang, Hien Nguyen, Raphael Sznitman, Yuan Xue
PublisherSpringer Science and Business Media Deutschland GmbH
Pages184-192
Number of pages9
ISBN (Print)9783030882099
DOIs
Publication statusPublished - 1 Jan 2021
Externally publishedYes
Event1st Workshop on Deep Generative Models for Medical Image Computing and Computer Assisted Intervention, DGM4MICCAI 2021 and 1st Workshop on Data Augmentation, Labelling, and Imperfections, DALI 2021 held in conjunction with 24th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2021 - Virtual, Online
Duration: 1 Oct 20211 Oct 2021

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13003 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference1st Workshop on Deep Generative Models for Medical Image Computing and Computer Assisted Intervention, DGM4MICCAI 2021 and 1st Workshop on Data Augmentation, Labelling, and Imperfections, DALI 2021 held in conjunction with 24th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2021
CityVirtual, Online
Period1/10/211/10/21

Keywords

  • Data augmentation
  • Latent space modelling
  • VAE

Fingerprint

Dive into the research topics of 'Data Augmentation with Variational Autoencoders and Manifold Sampling'. Together they form a unique fingerprint.

Cite this