TY - GEN
T1 - Data Augmentation with Variational Autoencoders and Manifold Sampling
AU - Chadebec, Clément
AU - Allassonnière, Stéphanie
N1 - Publisher Copyright:
© 2021, Springer Nature Switzerland AG.
PY - 2021/1/1
Y1 - 2021/1/1
N2 - 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.
AB - 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.
KW - Data augmentation
KW - Latent space modelling
KW - VAE
UR - https://www.scopus.com/pages/publications/85116893748
U2 - 10.1007/978-3-030-88210-5_17
DO - 10.1007/978-3-030-88210-5_17
M3 - Conference contribution
AN - SCOPUS:85116893748
SN - 9783030882099
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 184
EP - 192
BT - Deep Generative Models, and Data Augmentation, Labelling, and Imperfections - First Workshop, DGM4MICCAI 2021, and First Workshop, DALI 2021, Held in Conjunction with MICCAI 2021, Proceedings
A2 - Engelhardt, Sandy
A2 - Oksuz, Ilkay
A2 - Zhu, Dajiang
A2 - Yuan, Yixuan
A2 - Mukhopadhyay, Anirban
A2 - Heller, Nicholas
A2 - Huang, Sharon Xiaolei
A2 - Nguyen, Hien
A2 - Sznitman, Raphael
A2 - Xue, Yuan
PB - Springer Science and Business Media Deutschland GmbH
T2 - 1st 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
Y2 - 1 October 2021 through 1 October 2021
ER -