A Better Way to Pretrain Deep Boltzmann Machines

Abstract

We describe how the pre-training algorithm for Deep Boltzmann Machines (DBMs) is related to the pre-training algorithm for Deep Belief Networks and we show that under certain conditions, the pre-training procedure improves the variational lower bound of a two-hidden-layer DBM. Based on this analysis, we develop a different method of pre-training DBMs that distributes the modelling work more evenly over the hidden layers. Our results on the MNIST and NORB datasets demonstrate that the new pre-training algorithm allows us to learn better generative models.

Cite

Text

Hinton and Salakhutdinov. "A Better Way to Pretrain Deep Boltzmann Machines." Neural Information Processing Systems, 2012.

Markdown

[Hinton and Salakhutdinov. "A Better Way to Pretrain Deep Boltzmann Machines." Neural Information Processing Systems, 2012.](https://mlanthology.org/neurips/2012/hinton2012neurips-better/)

BibTeX

@inproceedings{hinton2012neurips-better,
  title     = {{A Better Way to Pretrain Deep Boltzmann Machines}},
  author    = {Hinton, Geoffrey E. and Salakhutdinov, Ruslan},
  booktitle = {Neural Information Processing Systems},
  year      = {2012},
  pages     = {2447-2455},
  url       = {https://mlanthology.org/neurips/2012/hinton2012neurips-better/}
}