Publica
Hier finden Sie wissenschaftliche Publikationen aus den FraunhoferInstituten. A new algorithm of nonGaussian component analysis with radial kernel functions
 Annals of the Institute of Statistical Mathematics 59 (2007), No.1, pp.5775 ISSN: 00203157 

 English 
 Journal Article 
 Fraunhofer FIRST () 
Abstract
We consider highdimensional data which contains a linear lowdimensional nonGaussian structure contaminated with Gaussian noise, and discuss a method to identify this nonGaussian subspace. For this problem, we provided in our previous work a very general semiparametric framework called nonGaussian component analysis (NGCA). NGCA has a uniform probabilistic bound on the error of finding the nonGaussian components and within this framework, we presented an efficient NGCA algorithm called Multiindex Projection Pursuit. The algorithm is justified as an extension of the ordinary projection pursuit (PP) methods and is shown to outperform PP particularly when the data has complicated nonGaussian structure. However, it turns out that multiindex PP is not optimal in the context of NGCA. In this article, we therefore develop an alternative algorithm called iterative metric adaptation for radial kernel functions (IMAK), which is theoretically better justifiable within the NGCA framework. We demonstrate that the new algorithm tends to outperform existing methods through numerical examples.