Skip to content

Latest commit

 

History

History
32 lines (22 loc) · 1.14 KB

README.md

File metadata and controls

32 lines (22 loc) · 1.14 KB
Raw

Spiking Neural Network - e-prop

Reimplementing the e-prop framework with jax and spyx.

Report: here

Dependencies

  • Python 3.12
  • spyx
  • tonic
  • nir
  • torchvision

File organization

  • lif_light.py: spyx implementation of neuron models
  • lif.py: basic adaptation of the original tensorflow code
  • autodiff.ipynb: verification of e-prop calculation
  • shd.ipynb: demo of e-prop on the SHD dataset
  • neuron_types.ipynb: playing around with neuron models
  • test_autodiff.ipynb: a few tests
  • utils.py: some useful methods

Original paper code repo: https://github.com/IGITUGraz/eligibility_propagation

Full paper: A solution to the learning dilemma for recurrent networks of spiking neurons. G Bellec*, F Scherr*, A Subramoney, E Hajek, Darjan Salaj, R Legenstein, W Maass

I copied part of the original code in the original_code folder, to check the results with jax. To run it, install tensorflow 2 (I adapted the code for compatibility).

Other implementations of e-prop can be find at https://github.com/YigitDemirag/eprop-jax (out of spyx framework, no autodiff) or https://github.com/ChFrenkel/eprop-PyTorch (pytorch).