Title:How to Round Subspaces: A New Spectral Clustering Algorithm

Abstract:Consider the problem of approximating a $k$-dimensional linear subspace with another {$k$-piecewise} constant subspace in spectral norm. This problem is at the heart of many spectral methods used in data clustering and graph partitioning. Our main contribution is a new spectral clustering algorithm: It can recover a $k$-partition such that the subspace corresponding to the span of its indicator vectors is $O(\sqrt{\mathrm{OPT}})$ close to the original subspace in spectral norm, with $\mathrm{OPT}$ being the minimum possible. Moreover our algorithm does not impose any restriction on the cluster sizes. Previously, no algorithm was known which could find a $k$-partition closer than $o(k \cdot \mathrm{OPT})$.
We present two applications for our algorithm. First one finds a disjoint union of $k$-expanders which approximate a given graph in spectral norm. The second one is for approximating the $k$-partition whose clusters have expansion at most $\varphi_k$ on graphs satisfying $\varphi_k \le O(\lambda_{k+1})$, where $\lambda_{k+1}$ is the $(k+1)^{th}$ eigenvalue of Laplacian matrix. This significantly improves upon the previous algorithms, which required $\varphi_k \le O(\lambda_{k+1}/k)$.