The Effects of Precision Constraints in a Backpropagation Learning Network

被引：44

作者：

Hollis, Paul W. ^{[1
]}

Harper, John S. ^{[1
]}

Paulos, John J. ^{[1
]}

机构：

[1] North Carolina State Univ, Dept Elect & Comp Engn, Raleigh, NC 27695 USA

来源：

NEURAL COMPUTATION | 1990年 / 2卷 / 03期

关键词：

D O I：

10.1162/neco.1990.2.3.363

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 presents a study of precision constraints imposed by a hybrid chip architecture with analog neurons and digital backpropagation calculations. Conversions between the analog and digital domains and weight storage restrictions impose precision limits on both analog and digital calculations. It is shown through simulations that a learning system of this nature can be implemented in spite of limited resolution in the analog circuits and using fixed point arithmetic to implement the backpropagation algorithm.

引用

页码：363 / 373

页数：11

共 7 条

[1]

[Anonymous], 1989, ADV NEURAL INFORM PR

[2]

Gilbert S. L., 1988, 810 MIT LINC LAB

[3] ARTIFICIAL NEURAL NETWORKS USING MOS ANALOG MULTIPLIERS [J].

HOLLIS, PW ;

PAULOS, JJ .

IEEE JOURNAL OF SOLID-STATE CIRCUITS, 1990, 25 (03) :849-855

[4]

Paulos J. J., 1988, 1988 IEEE International Symposium on Circuits and Systems. Proceedings (Cat. No.88CH2458-8), P499, DOI 10.1109/ISCAS.1988.14973

[5]

Paulos J. J., 1988, 1 ANN M INT NEUR NET

[6]

Rumelhart D. E., 1986, PARALLEL DISTRIBUTED, VI, P323

[7] NEW STRATEGIES FOR IMPROVING SPEECH ENHANCEMENT [J].

WHITE, MW ;

HOLDAWAY, RM ;

GUO, Y ;

PAULOS, JJ .

INTERNATIONAL JOURNAL OF BIO-MEDICAL COMPUTING, 1990, 25 (2-3) :101-124

← 1 →