Computer scienceData scienceMachine learningData preprocessingFeature engineering

Autoencoders

Theory

Bypassing the bottleneck

Report a typo

In the general case, we have two options for the hidden layer of the autoencoder:

  • we can make it smaller than the input layer;

  • we can make it greater than or equal to the length of the input layer.

What are the consequences of the second option?

Select one option from the list

Create a free account to access the full topic