Skip to content

A practical comparison between Hopfield Networks and Restricted Boltzmann Machines as content-addressable autoassociative memories.

License

Notifications You must be signed in to change notification settings

gbg141/bblr-hopfield-boltzmann

 
 

Repository files navigation

bblr-hopfield-boltzmann

The goal of this project is to compare the Hopfield Neural Network model and the Restricted Boltzmann Machine model when they are used as associative memories that can store pattern data sets.

The details about this project are published in our wiki.

Reproducibility

The default configured random seed produces the same exact results as the ones published in this paper.

Requirements

  • Python >= 2.7.0
  • NumPy >= 1.12.0
  • UNIX-like system shell.

Execution

Just execute the following command in the root of the project to generate data sets and test the models:

$ sh generate results.sh

This will create several JSON files in out/results/. When it finishes, execute the following command:

$ sh generate latex.sh > out/tables.tex

This will create a TeX file in out/tables.tex containing the result tables used in this paper.

About

A practical comparison between Hopfield Networks and Restricted Boltzmann Machines as content-addressable autoassociative memories.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 65.1%
  • TeX 34.6%
  • Shell 0.3%