Lecture 28 : Autoencoder

preview_player
Показать описание
Autoencoder, dimensionality reduction using neural network
Рекомендации по теме
Комментарии
Автор

Shouldn't the number of nodes in the hidden layer be d+1? He says MN+1.. Just needed clarification if I'm missing something.

PurtiRS
Автор

26:00 But even if I have random symmetrical weights in the symmetrical architecture, wouldn't the reconstruction be perfect ?

As we will effectively just multiply and then divide by some weights.

newbie
join shbcf.ru