Deep Belief Nets in C++ and CUDA C: Volume 2: Autoencoding in the Complex Domain

© Timothy Masters 2018

Timothy MastersDeep Belief Nets in C++ and CUDA C: Volume 2https://doi.org/10.1007/978-1-4842-3646-8_4

4. Autoencoding

Timothy Masters¹ 

(1)

Ithaca, New York, USA

 

Autoencoding as a means of creating deep belief nets predates the use of restricted Boltzmann machines (RBMs) . This is perhaps because of their conceptual simplicity. The most basic autoencoder is an ordinary feedforward network that has a single hidden layer and is trained to reproduce its inputs. It’s a prediction model in which the targets are the inputs. The idea is that if the hidden layer is in some sense relatively weak (perhaps by virtue of having few neurons or having limited weight magnitudes or some other form of regularization), this hidden layer will ...