A modified CMAC algorithm based on credit assignment

被引：31

作者：

Zhang, L ^{[1
]}

Cao, QX

Lee, J

Zhao, YZ

机构：

[1] Shanghai Jiao Tong Univ, Res Inst Robot, Shanghai 200030, Peoples R China

[2] Univ Wisconsin, Res Ctr Intelligent Maintenance Syst, Milwaukee, WI 53224 USA

来源：

NEURAL PROCESSING LETTERS | 2004年 / 20卷 / 01期

基金：

中国国家自然科学基金;

关键词：

CMAC; convergence property; credit assignment; learning interference;

D O I：

10.1023/B:NEPL.0000039430.04088.78

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

A Credit-Assignment CMAC (CA-CMAC) algorithm is proposed to reduce learning interference in conventional CMAC. In the proposed CA-CMAC, the error of the training sample distributed to the addressed memory cell is proportional to the cell's credibility, which is the inverse of the cell's activated times. The learning process of CA-CMAC is analyzed and conventional CMAC is proved to be a special case of CA-CMAC. Furthermore, the convergence properties of CA-CMAC both in batch learning and in incremental learning are investigated; meanwhile, the convergence theorems in the two learning schemes are obtained, respectively. Finally, simulations are carried out to testify the theorems and compare the performance of CA-CMAC with that of CMAC. Simulation results prove that CA-CMAC converges faster than conventional CMAC.

引用

页码：1 / 10

页数：10

共 15 条

[1]

[Anonymous], J DYN SYST MEAS CONT, DOI DOI 10.1115/1.3426922

[2] Dynamic control of a robot arm using CMAC neural networks [J].

Cembrano, G ;

Wells, G ;

Sarda, J ;

Ruggeri, A .

CONTROL ENGINEERING PRACTICE, 1997, 5 (04) :485-492

[3]

KWON S, 1990, THESIS LOUISIANA STA

[4] CMAC-BASED ADAPTIVE CRITIC SELF-LEARNING CONTROL [J].

LIN, CS ;

KIM, H .

IEEE TRANSACTIONS ON NEURAL NETWORKS, 1991, 2 (05) :530-533

[5] Learning convergence of CMAC technique [J].

Lin, CS ;

Chiang, CT .

IEEE TRANSACTIONS ON NEURAL NETWORKS, 1997, 8 (06) :1281-1292

[6] Enhance the performance of CMAC neural network via fuzzy theory and credit apportionment [J].

Lu, HC ;

Chang, JC .

PROCEEDING OF THE 2002 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS, VOLS 1-3, 2002, :715-720

[7] REAL-TIME DYNAMIC CONTROL OF AN INDUSTRIAL MANIPULATOR USING A NEURAL-NETWORK-BASED LEARNING CONTROLLER [J].

MILLER, WT ;

HEWES, RP ;

GLANZ, FH ;

KRAFT, LG .

IEEE TRANSACTIONS ON ROBOTICS AND AUTOMATION, 1990, 6 (01) :1-9

[8] A COMPARISON OF 5 ALGORITHMS FOR THE TRAINING OF CMAC MEMORIES FOR LEARNING CONTROL-SYSTEMS [J].

PARKS, PC ;

MILITZER, J .

AUTOMATICA, 1992, 28 (05) :1027-1035

[9] A Hybrid Maximum Error Algorithm with Neighborhood Training for CMAC [J].

Sayil, S ;

Lee, KY .

PROCEEDING OF THE 2002 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS, VOLS 1-3, 2002, :165-170

[10]

SHI R, 1996, MATRIX ANAL

← 1 2 →