xiatian815/DH-SNN

Codes for implementation of DH-SNNs

This code implements the DH-SNNs for various tasks. We select some typical training codes for tasks in the paper to present.

1. files and folders description:

• The pre-processing and training codes can be found in the folder that corresponds to the task. (The folder named "delayed_xor" and "multitimescale_xor" represent the self-designed delayed spiking XOR problem and multi-timescale spiking XOR problem, respectively.)

• The folder named "SNN_layers" contains the main codes for the implementation of DH-SNNs model.

2. The datasets:

   1. SHD and SSC datasets can be downloaded from https://zenkelab.org/resources/spiking-heidelberg-datasets-shd/
   2. GSC can be downloaded from https://tensorflow.google.cn/datasets/catalog/speech_commands/
   3. (P)S-MNIST: This dataset can be found in torchvision.datasets.MNIST
   4. DEAP can be downloaded from https://www.eecs.qmul.ac.uk/mmv/datasets/deap/
   5. TIMIT can be found here: https://catalog.ldc.upenn.edu/LDC93S1
   6. Self-designed delayed spiking XOR problem and multi-timescale spiking XOR problem.
   7. NeuroVPR task

3. Pre-requisites

• Python >=3.9
• HDF5 for python.
• Pytorch == 1.12.1, torchvision == 0.13.1, torchaudio == 0.12.1
• Preprocessing packages: [librosa](librosa — librosa 0.8.0 documentation),[tables](tables · PyPI),[wfdb](wfdb — wfdb 3.3.0 documentation),[klepto](klepto · PyPI) and [Scipy](SciPy.org — SciPy.org),
• matplotlib
• scikit-learn == 1.1.1
• pandas == 1.4.3
• pickle

4. Code running

• Data preprocessing.

  The datasets(SHD,SSC,GSC,TIMIT and DEAP) are required to arrange the data before training. The pre-processing codes and instructions can be found in the folder that corresponds to the task. The data of NeuroVPR is available on Zenodo: https://zenodo.org/records/7827108#.ZD_ke3bP0ds

• Model training. The training codes can be found in the folder that corresponds to the task. To start the training of DH-SNNs on SSC, for example, just go to the folder SSC and run

  # DH-SFNN on SSC
  python  main_dense_denri.py 
  

  or

  # DH-SRNN on SSC
  python  main_rnn_denri.py 
  

• Pre-trained models are provided for some tasks in the folder.