Comparative Study of Autoencoders-Its Types and Application

Autoencoder was first proposed by LeCun in 1987. Autoencoders are a type of artificial unsupervised neural network used to study data encodings. The purpose of the autoencoder is to get an excessive dimensional representation (code text) of high-resolution records. Dimensions are mainly reduced by training in a way that the final outcome focus on the important component. Autoencoders support us for the same. Autoencoders are designed and pre-programed neural network that rescale the input in a presentable form and back to original scale. This research study provides an overview of the different types of auto encoders, its features and areas, where it can be applied or implemented.