Ranking on Data Manifolds

Abstract

The Google search engine has enjoyed a huge success with its web\npage ranking algorithm, which exploits global, rather than local,\nhyperlink structure of the web using random walks. Here we propose\na simple universal ranking algorithm for data lying in the\nEuclidean space, such as text or image data. The core idea of our\nmethod is to rank the data with respect to the intrinsic manifold structure collectively revealed by a great amount of data.\nEncouraging experimental results from synthetic, image, and text\ndata illustrate the validity of our method.