SI_original_demo.ipynb : For artificial data experiments using original data(k is fixed)
SI_original_kchoice_demo.ipynb : For artificial data experiments using original data(k is adopted)
SI_original_bench_demo.ipynb : For benchmark data sets(k is fixed)
SI_latent_demo.ipynb : For experiments using latent data


Requirements
Python version:
	•	Python 3.10 or later
Required packages:
	•	numpy 1.26.4 or later
	•	torch 2.2.0 or later
	•	sicore 2.3.0 or later
Others
  pandas 
  matplotlib 
  matplotlib.pyplot
  seaborn
  sklearn 
  sklearn.preprocessing 
  collections
  math
  tqdm
  concurrent.futures
  scipy
  random
  csv
  itertools
  math
  sklearn 
  os
  torchvision
  PIL
  torchvision.utils
  scipy.spatial
  cv2
  glob

In benchmark dataset experiments and MVtec datasets, You should download  datasets which you would like to experiment.
All datasets are able to download in official web site or Kaggle.
