[NeurIPS 2024] Spiking Graph Neural Networks on Riemannian Manifolds

Get Started

Install Python packages.

pip install -r requirements.txt

Then you can run the command to train the model.

python main.py --task NC --dataset Physics --root_path ${your_path}

please replace ${your_path} with your dataset file path. If you want to use vallina SNN neurons, you can add --use_MS.

If you need to use product space of manifolds, you can add --use_product.

All the configuration of models can be load from Json file in ./configs.

Model Architecture

Results

Visualization

Figure 1. Visualization of 34-th node on KarateClub dataset on Torus manifold.

Figure 2. Visualization of 1-th node on KarateClub dataset on Sphere manifold.

Figure 3. Visualization of 17-th node on KarateClub dataset on Sphere manifold.