Development of Autoencoder and Variational Autoencoder for Image Recognition Using Convolutional Neural Network
摘要
In this work, we present the technology of developing the architecture of a standard and variational autoencoder based on a convolutional neural network and using the Keras library. We construct loss function plots for both the training and testing datasets and analyze the distribution of hidden layer vectors for both autoencoders. As a result of the research, conclusions are drawn about the advantages of using a convolutional neural network in autoencoder construction and the suitability of utilizing a variational autoencoder for image recognition.