Analog Implementation of a Novel Resistive-Type Sigmoidal Neuron

被引：71

作者：

Khodabandehloo, Golnar ^{[1
]}

Mirhassani, Mitra ^{[1
]}

Ahmadi, Majid ^{[1
]}

机构：

[1] Univ Windsor, Dept Elect & Comp Engn, Windsor, ON N9B 3P4, Canada

来源：

IEEE TRANSACTIONS ON VERY LARGE SCALE INTEGRATION (VLSI) SYSTEMS | 2012年 / 20卷 / 04期

关键词：

Activation function; analog neuron; sigmoid function; sigmoidal neuron; NETWORK; SENSITIVITY;

D O I：

10.1109/TVLSI.2011.2109404

中图分类号：

TP3 [计算技术、计算机技术];

学科分类号：

0812 ;

摘要：

An important part of any hardware implementation of artificial neural networks (ANNs) is realization of the activation function which serves as the output stage of each layer. In this work, a new NMOS/PMOS design is proposed for realizing the sigmoid function as the activation function. Transistors in the proposed neuron are biased using only one biasing voltage. By operating in both triode and saturation regions, the proposed neuron can provide an accurate approximation of the sigmoid function. The neuron circuit is designed and laid out in 90-nm CMOS technology. The proposed neuron can be potentially used in implementation of both analog and hybrid ANNs.

引用

页码：750 / 754

页数：6

共 23 条

[1] Approximation of sigmoid function and the derivative for hardware implementation of artificial neurons [J].

Basterretxea, K ;

Tarela, JM ;

del Campo, I .

IEE PROCEEDINGS-CIRCUITS DEVICES AND SYSTEMS, 2004, 151 (01) :18-24

[2] An experimental study on nonlinear function computation for neural/fuzzy hardware design [J].

Basterretxea, Koldo ;

Tarela, Jose Manuel ;

del Campo, Ines ;

Bosque, Guillermo .

IEEE TRANSACTIONS ON NEURAL NETWORKS, 2007, 18 (01) :266-283

[3] Sensitivity study and improvements on a nonlinear resistive-type neuron circuit [J].

Djahanshahi, H ;

Ahmadi, M ;

Jullien, GA ;

Miller, WC .

IEE PROCEEDINGS-CIRCUITS DEVICES AND SYSTEMS, 2000, 147 (04) :237-242

[4]

Djahanshahi H., 1998, ISCAS '98. Proceedings of the 1998 IEEE International Symposium on Circuits and Systems (Cat. No.98CH36187), P41, DOI 10.1109/ISCAS.1998.703891

[5] Quantization noise improvement in a hybrid distributed-neuron ANN architecture [J].

Djahanshahi, H ;

Ahmadi, M ;

Jullien, GA ;

Miller, WC .

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS II-ANALOG AND DIGITAL SIGNAL PROCESSING, 2001, 48 (09) :842-846

[6]

Djahanshahi H, 1996, IEEE IJCNN, P868, DOI 10.1109/ICNN.1996.549011

[7]

Djahanshahi H, 1996, ISCAS 96: 1996 IEEE INTERNATIONAL SYMPOSIUM ON CIRCUITS AND SYSTEMS - CIRCUITS AND SYSTEMS CONNECTING THE WORLD, VOL 3, P483, DOI 10.1109/ISCAS.1996.541638

[8] Scalable closed-boundary analog neural networks [J].

Fakhraie, SM ;

Farshbaf, H ;

Smith, KC .

IEEE TRANSACTIONS ON NEURAL NETWORKS, 2004, 15 (02) :492-504

[9] Analog neural network implementation for a real-time surface classification application [J].

Gatet, Laurent ;

Tap-Beteille, Helene ;

Lescure, Marc .

IEEE SENSORS JOURNAL, 2008, 8 (7-8) :1413-1421

[10] Comparison Between Analog and Digital Neural Network Implementations for Range-Finding Applications [J].

Gatet, Laurent ;

Tap-Beteille, Helene ;

Bony, Francis .

IEEE TRANSACTIONS ON NEURAL NETWORKS, 2009, 20 (03) :460-470

← 1 2 3 →