Spectral Methods for Cancer Classification Using Microarray Data

In this paper, we present a novel method based on spectral bipartitioning, traditionally used for finding min-cuts in graphs, for classification of cancer using microarray data. Our method is applied to five publicly available datasets of acute leukemia, colon cancer, ovarian cancer, prostate cancer and diffuse large B-cell lymphoma, and is shown to have classification accuracy comparable to that of some of the currently known best classification methods for microarray data.