Modified self-organizing feature map algorithms for efficient digital hardware implementation

被引:36
作者
Ienne, P
Thiran, P
Vassilas, N
机构
[1] SWISS FED INST TECHNOL, MICROCOMP LAB, CH-1015 LAUSANNE, SWITZERLAND
[2] DEMOKRITOS NATL CTR SCI RES, INST INFORMAT & TELECOMMUN, AGHIA PARASKEVI 15310, GREECE
来源
IEEE TRANSACTIONS ON NEURAL NETWORKS | 1997年 / 8卷 / 02期
关键词
Markov processes; neural-network hardware; self-organizing feature maps; systolic arrays;
D O I
10.1109/72.557669
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
This paper describes two variants of the Kohonen's self-organizing feature map (SOFM) algorithm, Both variants update the weights only after presentation of a group of input vectors, In contrast, in the original algorithm the weights are updated after presentation of every input vector, The main advantage of these variants is to make available a finer grain of parallelism, for implementation on machines with a very large number of processors, without compromising the desired properties of the algorithm, In this work it is proved that, for one-dimensional (1-D) maps and 1-D continuous input and weight spaces, the strictly increasing or decreasing weight configuration forms an absorbing class in both variants, exactly as in the original algorithm, Ordering of the maps and convergence to asymptotic values are also proved, again confirming the theoretical results obtained for the original algorithm, Simulations of a real-world application using two-dimensional (2-D) maps on 12-D speech data are presented to back up the theoretical results and show that the performance of one of the variants Is in all respects almost as good as the original algorithm, Finally, the practical utility of the finer parallelism made available is confirmed by the description of a massively parallel hardware system that makes effective use of the best variant.
引用
收藏
页码:315 / 330
页数:16
相关论文
共 24 条
[1]  
ALEKSANDER I, 1989, NEURAL COMPUTER ARCH
[2]  
[Anonymous], P 1990 INT JOINT C N
[3]  
[Anonymous], 1988, SELF ORG ASS MEMORY
[4]  
[Anonymous], 1990, APPL MATH
[5]   SELF-ORGANIZATION AND AS CONVERGENCE OF THE ONE-DIMENSIONAL KOHONEN ALGORITHM WITH NONUNIFORMLY DISTRIBUTED STIMULI [J].
BOUTON, C ;
PAGES, G .
STOCHASTIC PROCESSES AND THEIR APPLICATIONS, 1993, 47 (02) :249-274
[6]   CONSTRAINED TOPOLOGICAL MAPPING FOR NONPARAMETRIC REGRESSION-ANALYSIS [J].
CHERKASSKY, V ;
LARINAJAFI, H .
NEURAL NETWORKS, 1991, 4 (01) :27-40
[7]  
Clark D. S, 1978, STOCHASTIC APPROXIMA, V26
[8]  
Cornu T., 1996, Scientific Programming, V5, P47
[9]  
COTTRELL M, 1987, ANN I H POINCARE-PR, V23, P1
[10]   SELF-ORGANIZING MAPS - ORDERING, CONVERGENCE PROPERTIES AND ENERGY FUNCTIONS [J].
ERWIN, E ;
OBERMAYER, K ;
SCHULTEN, K .
BIOLOGICAL CYBERNETICS, 1992, 67 (01) :47-55