Thanks to visit codestin.com
Credit goes to github.com

Skip to content

kaitken17/mpn

 
 

Repository files navigation

Supporting code for Multi-Plasticity Networks.

To reproduce results from the paper, see the directory "paper".

The code contained here can be used to construct networks with mutli-plasticity layers, including the two-layer MPN analyzed in the paper. The notebook mpn_demonstration.ipynb demonstrates the following three architures:

  • The two-layer MPN analyzed in the paper.
  • A three-layer MPN (first two layers have multi-plasticity)
  • A recurrent MPN (input and recurrent layers have multi-plasticity).

LICENSING/CONTRIBUTING NOTE:

The following applies to files in this repo as well as the identically named files within the "paper" directory.

The LICENSING/CONTRIBUTING files in this repo only cover the following files:

  • analysis.py
  • context_data.py
  • mpn_demonstration.ipynb
  • paper/mpns_paper.ipynb

The following files inheret the associated LICENSING/CONTRIBUTING of the forked repo:

  • net_utils.py
  • networks.py

The following files inheret the associated LICENSING/CONTRIBUTING from: https://arxiv.org/pdf/2010.15114.pdf

  • int_data.py

About

Meta-learning local synaptic plasticity for continual familiarity detection

Resources

License

Contributing

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Jupyter Notebook 96.5%
  • Python 3.5%