Intuitive approach to DL

Hi all,

I’ve always been the sort of person who digs into intuition and tries to understand things on a fundamental motivation level. I wanna open this thread up so that we share intuitions that everyone uses, for example one could explain why variations of RNNs are first inceptionised for NLP, and so on. Let’s start sharing stuff here so that everyone is able to develop intuitions on what’s exactly motivating people to use certain techniques for certain applications.


If you want intuition on RNNs, you should take a look at Bayesian and markov networks. The idea of rnn was inspired by them