If you feel the other slides are TL;DR!
(I mean, aren't they interdependent and used together?)
That is what this talk will show you!
Feature selection, Dimensionality reduction
Also, exact algorithms, hyper-heuristics?
'restriction' in RBMs makes it easier to handle than normal Boltzmann machines, and thus less complex
As opposed to neural networks, RBM is only a 'two-layer' model if we may call it so (i.e., hidden and visible units comprising 2 layers independently)
Deep neural network (DNN) may be comprised of multiple hidden layers.
Also RBM, contrary to DNN, is an unsupervised model
Comparative study made with GLM, KNN, SVM and CNN - see table!