The anatomy of a large-scale hypertextual Web search engine
WWW7 Proceedings of the seventh international conference on World Wide Web 7
Semi-supervised graph clustering: a kernel approach
ICML '05 Proceedings of the 22nd international conference on Machine learning
A tutorial on spectral clustering
Statistics and Computing
ICDM '08 Proceedings of the 2008 Eighth IEEE International Conference on Data Mining
Fast approximate spectral clustering
Proceedings of the 15th ACM SIGKDD international conference on Knowledge discovery and data mining
IJCAI'03 Proceedings of the 18th international joint conference on Artificial intelligence
A regularized formulation for spectral clustering with pairwise constraints
IJCNN'09 Proceedings of the 2009 international joint conference on Neural Networks
Knowledge and Information Systems
Information Sciences: an International Journal
Mind the eigen-gap, or how to accelerate semi-supervised spectral learning algorithms
IJCAI'11 Proceedings of the Twenty-Second international joint conference on Artificial Intelligence - Volume Volume Three
Deflation-based power iteration clustering
Applied Intelligence
| Hi-index | 0.00 |
Spectral Clustering is a popular learning paradigm that employs the eigenvectors and eigenvalues of an appropriate input matrix for approximating the clustering objective. Albeit its empirical success in diverse application areas, spectral clustering has been criticized for its inefficiency when dealing with large-size datasets. This is mainly due to the fact that the complexity of most eigenvector algorithms is cubic with respect to the number of instances and even memory efficient iterative eigensolvers (such as the Power Method) may converge very slowly to the desired eigenvector solutions. In this paper, inspired from the relevant work on Pagerank we propose a semi-supervised framework for spectral clustering that provably improves the efficiency of the Power Method for computing the Spectral Clustering solution. The proposed method is extremely suitable for large and sparse matrices, where it is demonstrated to converge to the eigenvector solution with just a few Power Method iterations. The proposed framework reveals a novel perspective of semi-supervised spectral methods and demonstrates that the efficiency of spectral clustering can be enhanced not only by data compression but also by introducing the appropriate supervised bias to the input Laplacian matrix. Apart from the efficiency gains, the proposed framework is also demonstrated to improve the quality of the derived cluster models.