What is a Boltzmann machine?
A Boltzmann machine is a kind of recurrent neural network in which nodes make binary decisions. Boltzmann machines can be lined up to form more complex systems such as deep belief networks. A Boltzmann machine is also known as a stochastic Hopfield network with hidden units.
Although the Boltzmann machine was named after the Austrian scientist Ludwig Boltzmann who invented the Boltzmann distribution in the 20th century, this type of network was actually developed by Stanford scientist Geoff Hinton. It is closely related to the idea of a Hopfield network developed in the 1970s and draws on ideas from the world of thermodynamics to guide the work towards the desired states. Indeed, some experts might speak of certain types of Boltzmann machines as a 'stochastic Hopfield network with hidden units'.
In the Boltzmann machine there is a desire to achieve a 'thermal equilibrium' or to optimize the global energy distribution when the temperature and energy of the system are not literal but relative to the laws of thermodynamics. In one process, so-called simulated annealing, the Boltzmann machine, performs processes to slowly separate a large amount of noise from a signal. Boltzmann machines use stochastic binary units to balance the probability distribution, or in other words to minimize energy.
Boltzmann restricted machines are machines that have no interlayer connections in the hidden layers of the network.