Backpropagation with Biologically Plausible Spatio-Temporal Adjustment For Training Deep Spiking Neural Networks

This repository contains code from our paper [Backpropagation with Biologically Plausible Spatio-Temporal Adjustment For Training Deep Spiking Neural Networks]. If you use our code or refer to this project, please cite it using

@article{shenBackpropagationBiologicallyPlausible2021,
  title = {Backpropagation with {{Biologically Plausible Spatio-Temporal Adjustment For Training Deep Spiking Neural Networks}}},
  author = {Shen, Guobin and Zhao, Dongcheng and Zeng, Yi},
  year = {2021},
  month = oct,
  journal = {arXiv:2110.08858 [cs]},
  eprint = {2110.08858},
  eprinttype = {arxiv},
  primaryclass = {cs},
  archiveprefix = {arXiv} 
}

Requirments

• numpy
• scipy
• pytorch >= 1.7.0
• torchvision

Data preparation

First modify the DATA_DIR='path/to/datasets in code/datasets/__init__.py to the root directory of your datasets. Neuromorphic datasets NMNIST, DVS-Gesture and DVS-CIFAR10 need to be manually downloaded and placed under the /path/to/datasets/DVS/*

/path/to/datasets/
  DVS/
    DVS_Cifar10/
    DVS_Gesture/
    NMNIST/

Train

Run training scripts corresponding to different datasets.

For example, training and validating the proposed method on the MNIST dataset:

bash ./train_dvsg.sh