Generalization of iterative sampling in autoencoders - Archive ouverte HAL Access content directly
Preprints, Working Papers, ... Year :

Generalization of iterative sampling in autoencoders

(1) , (2) , (1) , (3) , (4) , (3)
1
2
3
4

Abstract

Generative autoencoders are designed to model a target distribution with the aim of generating samples and it has also been shown that specific non-generative autoencoders (i.e. contractive and denoising autoencoders) can be turned into gen-erative models using reinjections (i.e. iterative sampling). In this work, we provide mathematical evidence that any autoencoder reproducing the input data with a loss of information can sample from the training distribution using reinjections. More precisely, we prove that the property of modeling a given distribution and sampling from it not only applies to contractive and denoising autoencoders but also to all lossy autoencoders. In accordance with previous results, we emphasize that the reinjection sampling procedure in autoencoders improves the quality of the sampling. We experimentally illustrate the above property by generating synthetic data with non-generative autoencoders trained on standard datasets. We show that the learning curve of a classifier trained with synthetic data is similar to that of a classifier trained with original data.
Fichier principal
Vignette du fichier
ICMLA_IEEE.pdf (1010.73 Ko) Télécharger le fichier
Vignette du fichier
ICMLA_supplementary_materials.pdf (703.52 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

cea-02917445 , version 1 (19-08-2020)

Identifiers

  • HAL Id : cea-02917445 , version 1

Cite

Miguel Solinas, Clovis Galiez, Romain Cohendet, Stéphane Rousset, Marina Reyboz, et al.. Generalization of iterative sampling in autoencoders. 2020. ⟨cea-02917445⟩
262 View
307 Download

Share

Gmail Facebook Twitter LinkedIn More