bbhamux 0.2.0

Minimal library to construct Hierarchical Associative Memories

Barebones-HAMUX

HAMUX built using equinox, minimal implementation. A temporary solution as HAMUX is being rebuilt.

Build any hopfield network using energy fundamentals. See the original HAMUX documentation for explanation.

Install: File copy

All logic is in one file: hamux.py. Please copy this file into whatever project you are working on, and modify as needed. You will need to manually install dependencies:

pip install equinox jax
pip install pytest # for tests

Install correct version of jaxlib for your hardware (e.g., to run on GPUs).

Run demo.ipynb for an example training on MNIST. Works best with GPU

Install: Pip

Tests

All basic tests for this package are in test.py.

pytest test.py

Citation

If this repository is useful for this work, please cite the following:

@inproceedings{
hoover2022universal,
title={A Universal Abstraction for Hierarchical Hopfield Networks},
author={Benjamin Hoover and Duen Horng Chau and Hendrik Strobelt and Dmitry Krotov},
booktitle={The Symbiosis of Deep Learning and Differential Equations II},
year={2022},
url={https://openreview.net/forum?id=SAv3nhzNWhw}
}