Learning Spectral Clustering

Abstract

Spectral clustering refers to a class of techniques which rely on the eigen- structure of a similarity matrix to partition points into disjoint clusters with points in the same cluster having high similarity and points in dif- ferent clusters having low similarity. In this paper, we derive a new cost function for spectral clustering based on a measure of error between a given partition and a solution of the spectral relaxation of a minimum normalized cut problem. Minimizing this cost function with respect to the partition leads to a new spectral clustering algorithm. Minimizing with respect to the similarity matrix leads to an algorithm for learning the similarity matrix. We develop a tractable approximation of our cost function that is based on the power method of computing eigenvectors.

Cite

Text

Bach and Jordan. "Learning Spectral Clustering." Neural Information Processing Systems, 2003.

Markdown

[Bach and Jordan. "Learning Spectral Clustering." Neural Information Processing Systems, 2003.](https://mlanthology.org/neurips/2003/bach2003neurips-learning/)

BibTeX

@inproceedings{bach2003neurips-learning,
  title     = {{Learning Spectral Clustering}},
  author    = {Bach, Francis R. and Jordan, Michael I.},
  booktitle = {Neural Information Processing Systems},
  year      = {2003},
  pages     = {305-312},
  url       = {https://mlanthology.org/neurips/2003/bach2003neurips-learning/}
}