January 2018
Beginner to intermediate
284 pages
8h 35m
English
Boltzmann Machines (BMs) can be thought of as a particular form of log-linear Markov random field, for which the energy function is linear in its free parameters. To increase their representation capacity for complicated distributions, one can consider and increase the number of variables that are never observed, that is, hidden variables, or in this case, the hidden neurons. RBMs are built on top of BMs, in which the restrictions are applied to force no visible-visible and hidden-hidden connections.
Read now
Unlock full access