THE EFFECTS OF QUANTIZATION ON MULTILAYER NEURAL NETWORKS

被引：62

作者：

DUNDAR, G ^{[1
]}

ROSE, K ^{[1
]}

机构：

[1] RENSSELAER POLYTECH INST,CTR INTEGRATED ELECTR & ELECTR MFG,TROY,NY 12180

来源：

IEEE TRANSACTIONS ON NEURAL NETWORKS | 1995年 / 6卷 / 06期

关键词：

D O I：

10.1109/72.471364

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

The effect of weight quantization in multilayer neural networks is discussed. A method is derived by which one can predict the performance degradation at the output given the properties of the network and number of bits of quantization. Predictions from this method are evaluated against simulation results. An algorithm to decrease the noise at the output is presented and the results are compared with those above.

引用

页码：1446 / 1451

页数：6

共 14 条

[1]

DUNDAR G, 1992, 5TH P IEEE ASIC C RO, P419

[2]

DUNDAR G, IN PRESS NEURAL COMP

[3]

Holler M., 1989, P INT JOINT C NEURAL, P191

[4]

LEE BW, 1992, P IEEE ISCAS, P1545

[5] A DYNAMIC CMOS MULTIPLIER FOR ANALOG VLSI BASED ON EXPONENTIAL PULSE-DECAY MODULATION [J].

MASSENGILL, LW .

IEEE JOURNAL OF SOLID-STATE CIRCUITS, 1991, 26 (03) :268-276

[6]

MONTALVO AJ, 1992, P INT JOINT C NEURAL, P192

[7] AN ANALYSIS ON THE PERFORMANCE OF SILICON IMPLEMENTATIONS OF BACKPROPAGATION ALGORITHMS FOR ARTIFICIAL NEURAL NETWORKS [J].

REYNERI, LM ;

FILIPPI, E .

IEEE TRANSACTIONS ON COMPUTERS, 1991, 40 (12) :1380-1389

[8]

ROSSETTO O, 1989, IEEE MICRO MAG, P56

[9]

TELEAVEN P, 1989, IEEE MICRO MAG DEC, P8

[10]

TSIVIDIS Y, 1987, ELECTRON LETT, P1313

← 1 2 →