Paper Title
An Experimental Study on the Performance of Pattern Synthesis based on Variational Autoencoder

Abstract
This paper deals with the performance of pattern synthesis by variational autoencoder (VAE) that is a deep learning algorithm. For VAE learning, pattern synthesis, and performance evaluation, MNIST handwritten digit image database is used. From the deep neural network-based classification experiments using the synthesized patterns, we confirmed that VAE can be useful for pattern synthesis. Index Terms - Variational Autoencoder, Pattern Synthesis, Deep Neural Network, Pattern Classification.