Boltzmann Machine
A stochastic generative neural network that learns probability distributions over its inputs.
Overview
A Boltzmann Machine is a type of stochastic recurrent neural network that can learn to represent complex probability distributions over binary-valued data. Named after the Boltzmann distribution in statistical mechanics, it consists of visible units (inputs) and hidden units connected by symmetric weights.
Key Details
Restricted Boltzmann Machines (RBMs) simplify the architecture by removing connections within the same layer. RBMs were crucial in the deep learning revolution, used for pre-training deep networks layer by layer. They are applied in recommendation systems, dimensionality reduction, and collaborative filtering, though modern approaches have largely superseded them.