Realization of the Conscience Mechanism in CMOS Implementation of Winner-Takes-All Self-Organizing Neural Networks

This paper presents a complementary metal-oxide-semiconductor (CMOS) implementation of a conscience mechanism used to improve the effectiveness of learning in the winner-takes-all (WTA) artificial neural networks (ANNs) realized at the transistor level. This mechanism makes it possible to eliminate the effect of the so-called ¿dead neurons,¿ which do not take part in the learning phase competition. These neurons usually have a detrimental effect on the network performance, increasing the quantization error. The proposed mechanism comes as part of the analog implementation of the WTA neural networks (NNs) designed for applications to ultralow power portable diagnostic devices for online analysis of ECG biomedical signals. The study presents Matlab simulations of the network's model, discusses postlayout circuit level simulations and includes results of measurement completed for the physical realization of the circuit.

[1]  Witold Pedrycz,et al.  Adaptive Weight Change Mechanism for Kohonens's Neural Network Implemented in CMOS 0.18 um Technology , 2007, ESANN.

[2]  Rafal Dlugosz,et al.  Initialization mechanism in Kohonen neural network implemented in CMOS technology , 2008, ESANN.

[3]  Stanley C. Ahalt,et al.  Competitive learning algorithms for vector quantization , 1990, Neural Networks.

[4]  Michel Verleysen,et al.  Analog implementation of a Kohonen map with on-chip learning , 1993, IEEE Trans. Neural Networks.

[5]  Sied Mehdi Fakhraie,et al.  VLSI-Compatible Immplementations for Artificial Neural Networks , 1997 .

[6]  Carver Mead,et al.  Analog VLSI and neural systems , 1989 .

[7]  Willy Sansen,et al.  An easy-to-use mismatch model for the MOS transistor , 2002, IEEE J. Solid State Circuits.

[8]  Howard C. Card,et al.  Analog VLSI Circuits for Competitive Learning Networks , 1998 .

[9]  Marta Kolasa,et al.  Parallel asynchronous neighborhood mechanism for WTM Kohonen network implemented in CMOS technology , 2008, ESANN.

[10]  Chung-Yu Wu,et al.  A new analog implementation of the Kohonen neural network , 1993, 1993 International Symposium on VLSI Technology, Systems, and Applications Proceedings of Technical Papers.

[11]  Farokh B. Bastani,et al.  ANN with two-dendrite neurons and its weight initialization , 1992, [Proceedings 1992] IJCNN International Joint Conference on Neural Networks.

[12]  Gert Cauwenberghs,et al.  Learning on Silicon: Adaptive VLSI Neural Systems , 1999 .

[13]  Sied Mehdi Fakhraie,et al.  VLSI - Compatible Implementations for Artificial Neural Networks , 2012 .

[14]  R. Dlugosz,et al.  CMOS programmable asynchronous neighborhood mechanism for WTM kohonen neural network , 2008, 2008 15th International Conference on Mixed Design of Integrated Circuits and Systems.

[15]  M. V. Velzen,et al.  Self-organizing maps , 2007 .

[16]  Chip-Hong Chang,et al.  New adaptive color quantization method based on self-organizing maps , 2005, IEEE Transactions on Neural Networks.

[17]  Riccardo Rizzo,et al.  A comparison between habituation and conscience mechanism in self-organizing maps , 2006, IEEE Transactions on Neural Networks.

[18]  Laurent Gatet,et al.  Comparison Between Analog and Digital Neural Network Implementations for Range-Finding Applications , 2009, IEEE Transactions on Neural Networks.

[19]  Witold Pedrycz,et al.  Analog-Counter-Based Conscience Mechanism in Kohonen's Neural Network Implemented in CMOS 0.18 μm Technology , 2006, 2006 IEEE Workshop on Signal Processing Systems Design and Implementation.

[20]  David Zipser,et al.  Feature Discovery by Competive Learning , 1986, Cogn. Sci..

[21]  R. Wojtyna,et al.  Improved power-saving synapse for adaptive neuroprocessing on silicon , 2005 .

[22]  Panu Somervuo,et al.  Self-Organizing Maps and Learning Vector Quantization for Feature Sequences , 1999, Neural Processing Letters.

[23]  Hiroomi Hikawa.  Frequency-based multilayer neural network with on-chip learning , 1996, 1996 IEEE International Symposium on Circuits and Systems. Circuits and Systems Connecting the World. ISCAS 96.

[24]  José Luis Huertas,et al.  A CMOS analog adaptive BAM with on-chip learning and weight refreshing , 1993, IEEE Trans. Neural Networks.

[25]  T. Talaska,et al.  Current Mode Euclidean Distance Calculation Circuit for Kohonen's Neural Network Implemented in CMOS 0.18µm Technology , 2007, 2007 Canadian Conference on Electrical and Computer Engineering.

[26]  Shih-Lun Chen,et al.  A variable control system for wireless body sensor network , 2008, 2008 IEEE International Symposium on Circuits and Systems.

[27]  S. Tam,et al.  An electrically trainable artificial neural network (ETANN) with 10240 'floating gate' synapses , 1990, International 1989 Joint Conference on Neural Networks.

[28]  Rafal Dlugosz,et al.  Low power current-mode binary-tree asynchronous Min/Max circuit , 2010, Microelectron. J..

[29]  Duane DeSieno,et al.  Adding a conscience to competitive learning , 1988, IEEE 1988 International Conference on Neural Networks.

[30]  Bing J. Sheu,et al.  A programmable analog VLSI neural network processor for communication receivers , 1993, IEEE Trans. Neural Networks.

[31]  Wei Zhang,et al.  Physiological data acquisition system for education assessment using wireless sensor network , 2008, 2008 International Conference on Information Technology and Applications in Biomedicine.

[32]  Marwan A. Jabri,et al.  A hybrid analog and digital VLSI neural network for intracardiac morphology classification , 1995 .

[33]  M. Khalil Hani,et al.  ASIC design of a Kohonen neural network microchip , 2004, 2004 IEEE International Conference on Semiconductor Electronics.

[34]  R. Wojtyna,et al.  Experimental Kohonen neural network implemented in CMOS 0.18μm technology , 2008, 2008 15th International Conference on Mixed Design of Integrated Circuits and Systems.

[35]  Hiroomi Hikawa.  Frequency-based multilayer neural network with on-chip learning and enhanced neuron characteristics , 1999, IEEE Trans. Neural Networks.

[36]  Emile Fiesler,et al.  Neural Network Initialization , 1995, IWANN.