/rsc

Robust Spectral Clustering. Implementation of "Robust Spectral Clustering for Noisy Data: Modeling Sparse Corruptions Improves Latent Embeddings".

Primary LanguageJupyter NotebookMIT LicenseMIT

Robust Spectral Clustering (RSC)

Implementation of the method proposed in the paper: "Robust Spectral Clustering for Noisy Data: Modeling Sparse Corruptions Improves Latent Embeddings", Aleksandar Bojchevski, Yves Matkovic, and Stephan Günnemann, SIGKDD 2017.

Installation

python setup.py install

Requirements

  • numpy/scipy
  • sklearn

Demo

See example.ipynb for a comparison with vanilla Spectral Clustering on the moons dataset.

Cite

Please cite our paper if you use this code in your own work.

@inproceedings{bojchevski2017robust,
  title={Robust Spectral Clustering for Noisy Data: Modeling Sparse Corruptions Improves Latent Embeddings},
  author={Bojchevski, Aleksandar and Matkovic, Yves and G{\"u}nnemann, Stephan},
  booktitle={Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining},
  pages={737--746},
  year={2017},
  organization={ACM}
}