Autocodificadores Variacionales (VAE) Fundamentos Te\'oricos y Aplicaciones

TL;DR

This paper explains the theoretical foundations and applications of Variational Autoencoders (VAEs), a neural network-based probabilistic model for data encoding and generation, aimed at Spanish-speaking researchers.

Contribution

It provides a comprehensive theoretical overview of VAEs and discusses their practical applications, making this knowledge accessible to the Spanish-speaking scientific community.

Findings

Clarifies the probabilistic and neural network basis of VAEs

Describes how VAEs encode data in a latent space

Explains how VAEs generate new data samples

Abstract

VAEs are probabilistic graphical models based on neural networks that allow the coding of input data in a latent space formed by simpler probability distributions and the reconstruction, based on such latent variables, of the source data. After training, the reconstruction network, called decoder, is capable of generating new elements belonging to a close distribution, ideally equal to the original one. This article has been written in Spanish to facilitate the arrival of this scientific knowledge to the Spanish-speaking community.