Boltzmann machine

in its simplest form, a discrete time Hopfield network that em- ploys stochastic neurons and simulated an- nealing in its procedure for updating output values. More generally it can have hidden units and be subjected to supervised training so as to learn probabilities of different out- puts for each class of inputs.
Long
Cette description courte est en langue d'origine, la version longue traduite est en cours de préparation et sera disponible incéssament.