Skip to content

Latest commit

 

History

History
58 lines (51 loc) · 3.78 KB

README.md

File metadata and controls

58 lines (51 loc) · 3.78 KB
Raw

Python VersionLicenseMaintenance

NGC-Museum: Biomimetic Models with NGC-Learn

ngc-museum is a public repository for ngc-learn that houses biomimetic, brain-inspired computing, and computational neuroscience / biophysics models proposed throughout history. All models in this repo, whether contributed by community, other groups, or the ngc-learn dev team, are written in Python using ngc-learn (and JAX). Each model in the exhibits/ directory or collection of models in the exhibitors/ sub-directories contain README top-level files that explain their central properties and general organization of the sub-directory they are found within, including model/agent simulation instructions, problem task descriptions, as well as relevant hyper-parameter values need to reproduce experimental results.

For official walkthroughs going over the model exhibits found in this repo, please visit the ngc-learn documentation page: https://ngc-learn.readthedocs.io/ (under the "Model Museum" side-bar). For information, including anything related to usage instructions and details related to ngc-learn itself, please refer to the official ngc-learn repo (and its documentation).

For those contributing models/algorithms in either the exhibitors/ or exhibits/ directories, please send us an email if you are interested in writing your own walkthrough for us to include and integrate related to a particular model exhibit that you are working on in the official ngc-learn documentation as we warmly welcome the community to contribute to ngc-museum, as it is these contributions that help ensure various models of biomimetic inference/learning and brain-inspired computing see application as well as inspire future lines of scientific inquiry.

Current Models in the Museum

Models with Spiking Dynamics:

  1. Spiking neural network, trained with broadcast feedback alignment: Model, Walkthrough
  2. Diehl and Cook spiking network, trained with spike-timing-dependent plasticity (STDP): Model, Walkthrough
  3. Patch-level spiking network, trained with event-driven STDP: Model

Models with Graded Dynamics:

  1. Discriminative Predictive Coding: Model, Walkthrough
  2. Sparse coding (e.g., a Cauchy prior model & ISTA), trained with 2-factor Hebbian learning: Model, Walkthrough

This package is distributed under the 3-Clause BSD license.
It is currently maintained by the Neural Adaptive Computing (NAC) laboratory.