README

This package is a PyTorch port of the original Spiking Neural Networks with GoogLeNet-Like Inception Module(SpikeGoogle) framework for improved backpropagation based spiking neural networks (SNNs) learning with Inception modules. The original implementation is in C++ with CUDA and CUDNN.

Citation

Xuan Wang , Minghong Zhong , Hoiyuen Cheng, Junjie Xie, Yingchu Zhou, Jun Ren, Mengyuan Liu. "SpikeGoogle: Spiking Neural Networks with GoogLeNet-Like Inception Module." official paper version will be released soon.

What is this repository for?

• For learning synaptic weight and axonal delay parameters of a multilayer spiking neural network.

• Natively handles multiple spikes in each layer and error backpropagation through the layers.

• Implant inception inside the SNNs, where does SLAYER provide.

• Version 1.0

Requirements

Python 3 with the following packages installed:

• matplotlib==3.4.1
• numpy==1.20.2
• pickleshare==0.7.5
• PyYAML==5.4.1
• h5py==3.2.1
• torch==1.7.1
• torchvision==0.8.2

A CUDA enabled GPU is required for training any model. No plans on CPU only implementation yet. The software has been tested with CUDA libraries version 9.2 and GCC 7.3.0 on Ubuntu 18.04

Installation

The repository includes C++ and CUDA code that has to be compiled and installed before it can be used from Python, download the repository and run the following command to do so:

python setup.py install

To test the installation:

cd test
python -m unittest

Code

Any implementations can be found inside Examples folder.

Examples

• Run example CNN implementation

  cd  01_NMNIST_CNN
  tar -xvf NMNISTsmall.zip
  python nmnistCNN.py
  

Our main work

• Run SpikeGoogle implementation

  cd 03_NMNIST_3G
  tar -xvf NMNISTsmall.zip
  python nmnist3G.py
  

Contribution

• By Xuan Wang , Minghong Zhong , Hoiyuen Cheng, Junjie Xie, Yingchu Zhou, Jun Ren, Mengyuan Liu.

• This work builds on initial implementation by WangXuan[email protected].

Contact

For queries contact Wang Xuan