Spectral Clustering by Ellipsoid and Its Connection to Separable Nonnegative Matrix Factorization
This is an incremental improvement for researchers in clustering algorithms, addressing a known bottleneck in normalized cut methods.
The paper tackles the issue of sensitivity to initial points in spectral clustering by replacing K-means with a minimum volume enclosing ellipsoid, showing that this variant avoids poor clustering performance inherited from K-means.
This paper proposes a variant of the normalized cut algorithm for spectral clustering. Although the normalized cut algorithm applies the K-means algorithm to the eigenvectors of a normalized graph Laplacian for finding clusters, our algorithm instead uses a minimum volume enclosing ellipsoid for them. We show that the algorithm shares similarity with the ellipsoidal rounding algorithm for separable nonnegative matrix factorization. Our theoretical insight implies that the algorithm can serve as a bridge between spectral clustering and separable NMF. The K-means algorithm has the issues in that the choice of initial points affects the construction of clusters and certain choices result in poor clustering performance. The normalized cut algorithm inherits these issues since K-means is incorporated in it, whereas the algorithm proposed here does not. An empirical study is presented to examine the performance of the algorithm.