%0 Conference Proceedings %B Artificial Intelligence and Statistics %D 2015 %T Graph Approximation and Clustering on a Budget %A Ethan Fetaya %A Ohad Shamir %A Shimon Ullman %X
We consider the problem of learning from a similarity matrix (such as spectral cluster- ing and low-dimensional embedding), when computing pairwise similarities are costly, and only a limited number of entries can be observed. We provide a theoretical anal- ysis using standard notions of graph ap- proximation, significantly generalizing pre- vious results, which focused on spectral clustering with two clusters. We also pro- pose a new algorithmic approach based on adaptive sampling, which experimentally matches or improves on previous methods, while being considerably more general and computationally cheaper.
%B Artificial Intelligence and Statistics %V 38 %G eng