/Linear_GF

(PyTorch implementation) An expremely simple & fast graph filtering for CF

Primary LanguageJupyter Notebook

Graph Filter (linear)

  • Extremely faster implementation of linear low-pass graph filter part in GF-CF [Shen et al. CIKM'21]
  • Specifically, we use sparse matrices & GPU acceleration for the inference
  • You can find the original implementation of full GF-CF method in: https://github.com/yshenaw/GF_CF

Dependancy

  • numpy
  • Pytorch
  • Scipy

Notes

  • You can simply test & run cells in 'Graph Filtering (Light).ipynb'
  • Our code provides 'MovieLens-1M', 'Yelp', 'Gowalla', and 'Amazon-Book' datasets
  • One can possibly customize the code to implement batch inference-fashion based on your preference