A parallel neural processor for real-time applications

被引：5
作者：
Danese, G [1 ]
Leporati, F [1 ]
Ramat, S [1 ]
机构：
[1] Univ Pavia, INFM, I-27100 Pavia, Italy
来源：
IEEE MICRO | 2002年 / 22卷 / 03期
关键词：
D O I：
10.1109/MM.2002.1013301
中图分类号：
TP3 [计算技术、计算机技术];
学科分类号：
0812 ;
摘要：
Training and testing artificial neural networks can be challenging and time-consuming. Experiments with two real-time applications compared three approaches for implementing a multilayer perceptron neural network. As a result, Totem which is composed of special-purpose processor is said to be the most promising. This board demonstrated good recognition capability along with excellent computing times in executing the RTS algorithm, which turns out to be more affordable than BP in the absolute minimum search of a cost function.
引用
收藏
页码：20 / 31
页数：12
相关论文
共 20 条
[1]  
BAHILL AT, 1983, VISION RES, V23, P1573, DOI 10.1016/0042-6989(83)90171-2
[2]   TRAINING NEURAL NETS WITH THE REACTIVE TABU SEARCH [J].
BATTITI, R ;
TECCHIOLLI, G .
IEEE TRANSACTIONS ON NEURAL NETWORKS, 1995, 6 (05) :1185-1200
[3]   HARDWARE IMPLEMENTATION OF AN ARTIFICIAL NEURAL-NETWORK USING FIELD-PROGRAMMABLE GATE ARRAYS (FPGAS) [J].
BOTROS, NM ;
ABDULAZIZ, M .
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, 1994, 41 (06) :665-667
[4]  
DACIERNO A, 1993, P INT C ART NEUR NET, P1074
[5]   EXTREME VESTIBULO-OCULAR ADAPTATION INDUCED BY PROLONGED OPTICAL REVERSAL OF VISION [J].
GONSHOR, A ;
MELVILLJONES, G .
JOURNAL OF PHYSIOLOGY-LONDON, 1976, 256 (02) :381-414
[6]  
HAMMERSTROM D, 1990, P INT JOINT C NEUR N, P537
[7]  
Haykin S., 1994, NEURAL NETWORKS COMP
[8]   SPEEDING UP BACKPROPAGATION TRAINING ON A HYPERCUBE COMPUTER [J].
KERCKHOFFS, EJH ;
WEDMAN, FW ;
FRIETMAN, EEE .
NEUROCOMPUTING, 1992, 4 (1-2) :43-63
[9]  
LIU X, 1994, P INT C NEUR NETW, V1, P22
[10]   An efficient EM-based training algorithm for feedforward neural networks [J].
Ma, S ;
Ji, CY ;
Farmer, J .
NEURAL NETWORKS, 1997, 10 (02) :243-256
12