ML Anthology

Learning in Boltzmann Trees

Lawrence K. Saul, Michael I. Jordan
NeCo 1994 pp. 1174-1184
doi:10.1162/NECO.1994.6.6.1174 /neco/1994/saul1994neco-learning/

Abstract

We introduce a large family of Boltzmann machines that can be trained by standard gradient descent. The networks can have one or more layers of hidden units, with tree-like connectivity. We show how to implement the supervised learning algorithm for these Boltzmann machines exactly, without resort to simulated or mean-field annealing. The stochastic averages that yield the gradients in weight space are computed by the technique of decimation. We present results on the problems of N-bit parity and the detection of hidden symmetries.

NeCo Semantic Scholar

Cite

Text

Saul and Jordan. "Learning in Boltzmann Trees." Neural Computation, 1994. doi:10.1162/NECO.1994.6.6.1174

Markdown

[Saul and Jordan. "Learning in Boltzmann Trees." Neural Computation, 1994.](https://mlanthology.org/neco/1994/saul1994neco-learning/) doi:10.1162/NECO.1994.6.6.1174

BibTeX

@article{saul1994neco-learning,
  title     = {{Learning in Boltzmann Trees}},
  author    = {Saul, Lawrence K. and Jordan, Michael I.},
  journal   = {Neural Computation},
  year      = {1994},
  pages     = {1174-1184},
  doi       = {10.1162/NECO.1994.6.6.1174},
  volume    = {6},
  url       = {https://mlanthology.org/neco/1994/saul1994neco-learning/}
}