Spectral K-Way Ratio-Cut Partitioning and Clustering

Recent research on partitioning has focussed on the ratio-cut cost metric which maintains a balance between the sizes of the edges cut and the sizes of the partitions without fixing the size of the partitions a priori. Iterative approaches and spectral approaches to two-way ratio-cut partitioning have yielded higher quality partitioning results. In this paper we develop a spectral approach to multiway ratio-cut partitioning which provides a generalization of the ratio-cut cost metric to k-way partitioning and a lower bound on this cost metric. Our approach involves finding the k smallest eigenvalue/eigenvector pairs of the Laplacian of the graph. The eigenvectors provide an embedding of the graph's n vertices into a k-dimensional subspace. We devise a time and space efficient clustering heuristic to coerce the points in the embedding into k partitions. Advancement over the current work is evidenced by the results of experiments on the standard benchmarks.