Alternating Minimization and Boltzmann Machine Learning

Download: PDF.

“Alternating Minimization and Boltzmann Machine Learning” by W. Byrne. IEEE Transactions on Neural Networks, vol. 3, no. 4, 1992, pp. 612-620 (9 pages).

Abstract

Training a Boltzmann machine with hidden units is appropriately treated in information geometry using the information divergence and the technique of alternating minimization. The resulting algorithm is shown to be closely related to gradient descent Boltzmann machine learning rules, and the close relationship of both to the EM algorithm is described. An iterative proportional fitting procedure is described and incorporated into the alternating minimization algorithm.

Download: PDF.

BibTeX entry:

@article{byrne:bmalg,
   author = {W. Byrne},
   title = {Alternating {M}inimization and {B}oltzmann {M}achine Learning},
   journal = {IEEE Transactions on Neural Networks},
   volume = {3},
   number = {4},
   pages = {612--620 (9 pages)},
   year = {1992}
}

Back to Bill Byrne publications.