Correntropy: properties and applications in non-gaussian signal processing

被引：1281

作者：

Liu, Weifeng ^{[1
]}

Pokharel, Puskal P. ^{[1
]}

Principe, Jose C. ^{[1
]}

机构：

[1] Univ Florida, Dept Elect & Comp Engn, Gainesville, FL 32611 USA

来源：

IEEE TRANSACTIONS ON SIGNAL PROCESSING | 2007年 / 55卷 / 11期

基金：

美国国家科学基金会;

关键词：

generalized correlation function; information theoretic learning; kernel methods; metric; temporal principal component analysis (TPCA);

D O I：

10.1109/TSP.2007.896065

中图分类号：

TM [电工技术]; TN [电子技术、通信技术];

学科分类号：

0808 ; 0809 ;

摘要：

The optimality of second-order statistics depends heavily on the assumption of Gaussianity. In this paper, we elucidate further the probabilistic and geometric meaning of the recently defined correntropy function as a localized similarity measure. A close relationship between correntropy and M-estimation is established. Connections and differences between correntropy and kernel methods are presented. As such correntropy has vastly different properties compared with second-order statistics that can be very useful in non-Gaussian signal processing, especially in the impulsive noise environment. Examples are presented to illustrate the technique.

引用

页码：5286 / 5298

页数：13

共 25 条

[1] [Anonymous], INT CONF ACOUST SPEE
[2] THEORY OF REPRODUCING KERNELS
ARONSZAJN, N
[J]. TRANSACTIONS OF THE AMERICAN MATHEMATICAL SOCIETY, 1950, 68 (MAY) : 337 - 404
[3] Cherkassky V, 1997, IEEE Trans Neural Netw, V8, P1564, DOI 10.1109/TNN.1997.641482
[4] DuMouchel W., 1991, COMPUTING GRAPHICS S, P41, DOI 10.1007/978-1-4613-9154-8_3
[5] Beyond second-order statistics for learning: A pairwise interaction model for entropy estimation
Deniz Erdogmus
Jose C. Principe
Kenneth E. Hild
[J]. Natural Computing, 2002, 1 (1) : 85 - 108
[6] A mutual information extension to the matched filter
Erdogmus, D
Agrawal, R
Principe, JC
[J]. SIGNAL PROCESSING, 2005, 85 (05) : 927 - 935
[7] Generalized information potential criterion for adaptive system training
Erdogmus, D
Principe, JC
[J]. IEEE TRANSACTIONS ON NEURAL NETWORKS, 2002, 13 (05): : 1035 - 1044
[8] An error-entropy minimization algorithm for supervised training of nonlinear adaptive systems
Erdogmus, D
Principe, JC
[J]. IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2002, 50 (07) : 1780 - 1786
[9] Gretton A, 2005, J MACH LEARN RES, V6, P2075
[10] Feature extraction using information-theoretic learning
Hild, Kenneth E., II
Erdogmus, Deniz
Torkkola, Kari
Principe, Jose C.
[J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2006, 28 (09) : 1385 - 1392

← 1 2 3 →