Modern Deep Learning for Tabular Data: Novel Approaches to Common Modeling Problems

© The Author(s), under exclusive license to APress Media, LLC, part of Springer Nature 2023

A. Ye, Z. WangModern Deep Learning for Tabular Datahttps://doi.org/10.1007/978-1-4842-8692-0_8

8. Autoencoders

Andre Ye¹   and Zian Wang²

(1)

Seattle, WA, USA

(2)

Redmond, WA, USA

 

Weak encoding means mistakes and weak decoding means illiteracy.

—Rajesh Walecha, Author

An autoencoder is a very simple model: a model that predicts its own input. In fact, it may seem deceivingly simple to the point of being worthless. (After all, what use is a model that predicts what we already know?) Autoencoders are extraordinarily valuable and versatile architectures not because of their functionality to reproduce the input, but because of the internal capabilities developed in ...