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

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.