Learning Highly Structured Manifolds: Harnessing the Power of SOMs

Abstract. In this paper we elaborate on the challenges of learning manifolds that have many relevant clusters, and where the clusters can have widely varying statistics. We call such data manifolds highly structured. We describe approaches to structure identification through self-organized learning, in the context of such data. We present some of our recently developed methods to show that self-organizing neural maps contain a great deal of information that can be unleashed and put to use to achieve detailed and accurate learning of highly structured manifolds, and we also offer some comparisons with existing clustering methods on real data. 1 The Challenges of Learning Highly Structured Manifolds Data collected today are often high-dimensional due to the vast number of attributes that are of interest for a given problem, and which advanced instrumentation and computerized systems are capable of acquiring and managing. Owing to the large number of attributes that are designed to provid...

• Type: Conference
• Year: 2009
• Where: DAGSTUHL
• Authors: Erzsébet Merényi, Kadim Tasdemir, Lili Zhang
• Keywords: DAGSTUHL 2007, DAGSTUHL 2009, Hyperspectral Imagery, Manifolds, Relevant Clusters