Development of a variational autoencoder for handwritten digit recognition

Abstract

With each passing day, the application of neural networks becomes increasingly noticeable in various fields of activity. Scientists and researchers strive to develop new and improve existing neural networks to address issues in forecasting, creative tasks, medicine, and particularly image recognition. An essential tool in achieving this goal is variational autoencoders. A Variational Autoencoder (VAE) is a type of neural network applied for encoding data into a latent space. The latent space is a space with lower dimensionality than the data space.