Boltzmann Machines : High-Order Interactions and Synchronous Learning
A now classical innovative paper [H.S.A.] by Hinton-Sejnowski-Ackley introduced a class of formal neural networks, the Boltzmann machines, governed by asynchronous stochastic dynamics, quadratic energy functions, and pairwise interactions defined by synaptic weights. One of the exciting aspects of [H.S.A.] was the derivation of a locally implementable learning rule linked to a scheme of decreasing (artificial) temperatures, in the spirit of simulated annealing.
KeywordsLearning Rule Gibbs Distribution Stochastic Network Boltzmann Machine Random Configuration
Unable to display preview. Download preview PDF.
- [A]R. Azencott - - Markov fields and low-level vision tasks, Proc. Int. Cong. App. Math. ICIAM, Paris (1987).Google Scholar
- R. Azencott Gibbs fields, simulated annealing, and low-level vision tasks, Proc. Cong. Pattern Recognition, AFCET-INRIA, Antibes (1987).Google Scholar
- R. Azencott Synchronous Boltzmann machines : learning rules,Proc. Congress “Neural networks”, Les Arcs (1989), Springer-Verlag, NATO series (1990), vol. 68, Editors : Fogelman-Herault.Google Scholar
- R. AzencottParameter estimation for synchronous Markov fields (to appear).Google Scholar
- [Bo]P. Bourlard- Multilayer perceptions and learning, Proc. Congress “Neural networks”, Les Arcs (1989), to appear in Springer-Verlag, NATO series (1990).Google Scholar
- [G.G.]D. and S. Geman - Gibbs fields, simulated annealing, and Bayesian reconstruction of images, IEEE, PAMI (1984).Google Scholar
- [H.S.A.]G. Hinton, T. Sejnowski, D.H. Ackley - Boltzmann machines: constraint satisfaction networks that learn, Technical Report, Carnegie Mellon University (1984).Google Scholar