Resumen de Double-Layer Stacked Denoising Autoencoders for Regression
María Elena Fernández García, Antonio Ros Serrano, Eloy Hontoria Hernández, Aníbal Ramón Figueiras Vidal, José Luis Sancho Gómez
A Stacked Denoising Autoencoder (SDAE) is a deep neural network (NN) model trained and designed in one-by-one stacked layers to reconstruct the non-noisy version of the original input data. It is an architecture used with great success in statistical pattern recognition problems.The objective of this contribution is to determine if an MSDAE can benefit from the greater capabilities of representation of features obtained when two layers are introduced in the stacking process instead of a single one. To do this, the design and performance of these machines in regression problems are presented and analyzed both in terms of error and calculation cost.The experimental results underline interesting performance capabilities for specific purposes.
