Autoencoders are a special type of neural network where you have the same number of input and output neurons. The AutoEncoder learns to pass the data through to the output neurons after compressing it to a limited number of hidden neurons. This compression is a form of feature reduction. Denoising auto encoders can even be trained to remove noise.
Code for This Video:
https://github.com/jeffheaton/t81_558...
Course Homepage: https://sites.wustl.edu/jeffheaton/t8...
Follow Me/Subscribe:
/ heatonresearch
https://github.com/jeffheaton
/ jeffheaton
Support Me on Patreon: / jeffheaton