/neuralkernel

neural networks as a general-purpose computational framework

Primary LanguagePythonMIT LicenseMIT

Neuralkernel

Build Status PyPI PyPI - License PyPI - Python Version

This project uses networks of neuron-like computational units to build a framework of computation. Specifically, it implements characteristics traditionally found in neural networks including synaptic diversity, temporal delays, and voltage spikes. It builds on the ideas proposed in the paper STICK: Spike Time Interval Computational Kernel, A Framework for General Purpose Computation.

Getting Started

To run a sample network, you can run the module.

python -m neuralkernel

The networks currently implemented are:

  • Inverting Memory
  • Logarithm
  • Maximum
  • Non-Inverting Memory
  • Full Subtractor

For more information on each of these networks, please check out the docs folder.

Running the tests

To run the unit tests, you can run the following.

pytest