STOCHASTIC INFORMATION GRADIENT ALGORITHM WITH GENERALIZED GAUSSIAN DISTRIBUTION MODEL

被引：21
作者：
Chen, Badong [1 ]
Principe, Jose C. [1 ]
Hu, Jinchun [2 ]
Zhu, Yu [2 ]
机构：
[1] Univ Florida, Dept Elect & Comp Engn, Gainesville, FL 32611 USA
[2] Tsinghua Univ, Dept Precis Instruments & Mechanol, Beijing 100084, Peoples R China
来源：
JOURNAL OF CIRCUITS SYSTEMS AND COMPUTERS | 2012年 / 21卷 / 01期
基金：
美国国家科学基金会; 中国国家自然科学基金;
关键词：
Minimum error entropy criterion (MEE); stochastic information gradient (SIG) algorithm; generalized Gaussian density (GGD); least mean p-power (LMP); POWER ERROR CRITERION; ADAPTIVE ALGORITHM; PARAMETER;
D O I：
10.1142/S0218126612500065
中图分类号：
TP3 [计算技术、计算机技术];
学科分类号：
0812 ;
摘要：
This paper presents a parameterized version of the stochastic information gradient (SIG) algorithm, in which the error distribution is modeled by generalized Gaussian density (GGD), with location, shape, and dispersion parameters. Compared with the kernel-based SIG (SIG-Kernel) algorithm, the GGD-based SIG (SIG-GGD) algorithm does not involve kernel width selection. If the error is zero-mean, the SIG-GGD algorithm will become the least mean p-power (LMP) algorithm with adaptive order and variable step-size. Due to its well matched density estimation and automatic switching capability, the proposed algorithm is favorably in line with existing algorithms.
引用
收藏
页数：16
相关论文
共 24 条
[1]  
Abramowitz M., 1972, HDB MATH FUNCTIONS
[2]  
[Anonymous], 1985, Adaptive signal processing prentice-hall
[3]   Stochastic gradient algorithm under (h,φ)-entropy criterion [J].
Chen, B. ;
Hu, J. ;
Pu, L. ;
Sun, Z. .
CIRCUITS SYSTEMS AND SIGNAL PROCESSING, 2007, 26 (06) :941-960
[4]   Flexible independent component analysis [J].
Choi, S ;
Cichocki, A ;
Amari, SI .
JOURNAL OF VLSI SIGNAL PROCESSING SYSTEMS FOR SIGNAL IMAGE AND VIDEO TECHNOLOGY, 2000, 26 (1-2) :25-38
[5]  
Cover T.M., 1991, ELEMENT INFORM THEOR
[6]   STOCHASTIC GRADIENT ADAPTATION UNDER GENERAL ERROR CRITERIA [J].
DOUGLAS, SC ;
MENG, THY .
IEEE TRANSACTIONS ON SIGNAL PROCESSING, 1994, 42 (06) :1335-1351
[7]  
Erdogmus D., 2000, Second International Workshop on Independent Component Analysis and Blind Signal Separation. Proceedings, P75
[8]   Online entropy manipulation: Stochastic information gradient [J].
Erdogmus, D ;
Hild, KE ;
Principe, JC .
IEEE SIGNAL PROCESSING LETTERS, 2003, 10 (08) :242-245
[9]   Convergence properties and data efficiency of the minimum error entropy criterion in adaline training [J].
Erdogmus, D ;
Principe, JC .
IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2003, 51 (07) :1966-1978
[10]   Generalized information potential criterion for adaptive system training [J].
Erdogmus, D ;
Principe, JC .
IEEE TRANSACTIONS ON NEURAL NETWORKS, 2002, 13 (05) :1035-1044
123