Generalized Mahalanobis depth in the reproducing kernel Hilbert space
Tóm tắt
In this paper, Mahalanobis depth (MHD) in the Reproducing Kernel Hilbert Space (RKHS) is proposed. First, we extend the notion of MHD to a generalized version, i.e., the generalized MHD (GHMD), to make it suitable for the small sample with singular covariance matrix. We prove that GMHD is consistent with MHD when the sample has a full-rank covariance matrix. Second, we further extend GMHD to RKHS, i.e, the kernel mapped GMHD (kmGMHD), and discuss its main properties. Numeric results show that kmGMHD can give a better depth interpretation for the sample with special shape, such as a non-convex sample set. Our proposed kmGMHD can be potentially used as a robust tool for outliers detection and data classification. In addition, we also discuss the influence of parameters on the shape of the central regions.
Tài liệu tham khảo
Abe S (2005) Support vector machines for pattern classification, 1st edn. Springer-Verlag London Limited, Reading, pp 32–33 subsection 2.3
Canu S, Grandvalet Y, Guigue V, Rakotomamonjy A (2005) Svm and kernel methods matlab toolbox. Perception Systèmes et Information, INSA de Rouen, Rouen, France
Chen Z, Tyler DE (2004) On the finite sample breakdown points of redescending m-estimates of location. Stat Probab Lett 69: 233–242
Chenouri S (2004) Multivariate robust nonparametric inference based on data depth. Univ Waterloo 90: 67–89
Cui X, Lin L, Yang GR (2008) An extended projection data depth and its applications to discrimination. Commun Stat Theory Methods 37(14): 2276–2290
Donoho DL (1982) Breakdown properties of multivariate location estimators. PhD thesis, Deptment of Statistics, Harvard University
Gao Y (2003) Data depth based on spatial rank. Stat Probab Lett 65: 217–225
Ghosh AK, Chaudhuri P (2005a) On data depth and distribution-free discriminant analysis using separating surfaces. Bernoulli 11: 1–27
Ghosh AK, Chaudhuri P (2005b) On maximum depth and related classifiers. Board Found Scand Stat 32: 327–350
Hofmann T, SchöLkopf B, Smola AJ (2008) Kernel methods in machine learning. Ann Stat 36(3): 1171–1220
Jörnsten R (2004) Clustering and classification based on the l 1 data depth. J Multivar Anal 90: 67–89
Li G, Zhang J (1998) Sphering and its properties. Sankhyā Indian J Stat 60(Series A, Pt. 1): 119–133
Liu R, Singh K (1993) A quality index based on data depthand multivariate rank tests. J Am Stat Assoc 88: 252–260
Liu R, Singh K (1997) Notionso of limiting p values based on data depth and bootstrap. J Am Stat Assoc 92: 266–277
Liu RY (1990) On a notion of data depth based on random simplices. Ann Stat 18: 405–414
Liu RY (1992) Data depth and multivariate rank tests. In: Dodge Y (eds) L 1-Statistics and related methods. North-Holland, Amsterdam, pp 279–294
Mahalaobis PC (1936) On the generalized distance in statistics. Proc Natl Acad Sci India 12: 49–55
Pennacchi P (2008) Robust estimate of excitations in mechanical systems using m-estimators. J Sound Vib 310: 923–946
Stahel WA (1981) Robuste schatzungen: Infintesimale optimalitat and schatzungen von kovarianzmatrizen. PhD thesis, Zurich
Taylor JS, Cristianini N (2004) Kernel methods for pattern analysis, 1st edn. Cambridge University Press, Reading
Tian X, Vardi Y, Zhang C (2002) l 1−depth, depth realtive to a model, and robust regression. In: Dodge Y (eds) In Statistical data analysis based on the L 1−norm and Related Methods. Birkhauser, Basel, pp 285–299
Tukey JW (1975) Mathematics and picturing of data. In: James RD (eds) Proceedings of the international congress on mathematics, vol 2. Canadian Mathematics Congress, Vancouver, pp 523–531
Vardi Y, Zhang C (2000) The multivariate l 1-median and associated data depth. In: Proceedings of National Academy of Science, vol 97, pp 1423–1426
Zuo Y, Serfling R (2000) General notions of statistical depth function. Ann Stat 28(2): 461–482