Iterative Spectral Method (ISM) was originally proposed in the paper Iterative Spectral Method for Alternative Clustering, Wu et al., AISTATS 2018. It is an optimization technique that could be used to solve many HSIC related problems with a Gaussian Kernel. This repository contains an implemented ISM for Kernel Dimensionality Alternative Clustering (KDAC). KDAC was originally proposed in the paper Iterative Discovery from Multiple Clustering Views, Niu et al., Transactions on Pattern Analysis and Machine Intelligence, 2014.

The code is written with python2.7 on a Mint Linux Machine version 17.1. The following libraries must be included.

1. numpy
2. matplotlib
3. sklearn
4. scipy

Note: The GPU version is current NOT supported in this release.

The experiments should run by uncommenting the appropriate experiment within main.py and running the file.

This project is licensed under the MIT License - see the LICENSE.md file for details

If you intend to use our code in your research, please cite our paper as follows:

@inproceedings{wu2017ISM,
  title={Iterative Spectral Method for Alternative Clustering},
  author={Wu, Chieh and Ioannidis, Stratis and Mario, Sznaier and Xiangyu, Li and David, Kaeli and Jennifer, Dy},
  booktitle={Artificial Intelligence and Statistics},
  year={2018}
}

• Chieh Wu , Stratis Ioannidis , Mario Sznaier , Xiangyu Li , Yale Chang , David Kaeli , Jennifer Dy

We would like to acknowledge support for this project from the NSF grant IIS-1546428.