Visual-Explanations-from-Spiking-Neural-Networks-using-Interspike-Intervals

Kim, Y. & Panda, P., Visual explanations from spiking neural networks using inter-spike intervals. Sci Rep 11, 19037 (2021). https://doi.org/10.1038/s41598-021-98448-0

Prerequisites

• Python 3.9
• PyTorch 1.10.0
• NVIDIA GPU (>= 12GB)
• CUDA 10.2 (optional)

Getting Started

Conda Environment Setting

conda create -n VisualExp 
conda activate VisualExp
conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch

Training and testing

Training

• Download tinyimagenet dataset to PATH/TO/DATASET
• Train a model using BNTT (https://github.com/Intelligent-Computing-Lab-Yale/BNTT-Batch-Normalization-Through-Time).

Testing (on pretrained model)

• As a first step, download pretrained parameters (link) to PATH/TO/MODEL.pth.tar

• The above pretrained model is for TinyImageNet / VGG11 architecture

• Run the following command

python main.py --pretrainedmodel_pth 'PATH/TO/MODEL' --dataset_pth 'PATH/TO/DATASET' --target_layer 6

• Heatmaps (across timesteps) are visualized in folder figuresave

• In order to change the target images, please change the list in line 32.

  Example) visualize 10th and 52th images in TinyImageNet validation dataset.

img_nums = [10, 52]

Citation

Please consider citing our paper:

@article{kim2021visual,
 title={Visual explanations from spiking neural networks using interspike intervals},
 author={Kim, Youngeun and Panda, Priyadarshini},
 journal={arXiv preprint arXiv:2103.14441},
 year={2021}