Unipolar Input Signals in Single-Layer Feed-Forward Neural Networks

  • Anne-Johan Annema
Part of the The Springer International Series in Engineering and Computer Science book series (SECS, volume 314)

Abstract

In neural networks, signal processing is in principle performed by simple processors (neurons) operating in parallel. Implementing neural networks in analog hardware seems therefore natural. In the aim at reducing the chip size and the power consumption, which requires simplification of the signal processing blocks, the convergence of the neural network generally slows down. In analog hardware implementations of neural networks, two quadrant multipliers have several advantages over four quadrant multipliers:
  • two-quadrant multipliers (including their drivers) generally have a lower circuit complexity

  • two-quadrant multipliers generally have a lower power consumption than equally fast four quadrant multipliers

Keywords

eliO 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. [1]
    S. Bibyk and MIsmail, “Issues in Analog VLSI and MOS Techniques for Neural Computing”, In (C. Mead and MIsmail, eds.) Analog VLSI Implementation of Neural System, Kluwer, 103–133, 1989Google Scholar
  2. [2]
    K. Bult and H. Wallinga, “A CMOS four-quadrant analog multiplier”, IEEE Journal of Solid-State Circuits, vol. 21, 430–435, 1986CrossRefGoogle Scholar
  3. [3]
    H.A Castro, S.M. Tam and M.A. Holler, “Implementation and Performance of an Analog Nonvolatile Neural Network”, Analog Integrated Circuits and Signal Processing, vol. 4, 97–113, 1993CrossRefGoogle Scholar
  4. [4]
    D.D. Caviglia, M. Valle and G.M. Bisio, “Effects of Weight Discretization on the Back Propagation Learning Method: Algorithm Design and Hardware realization”, In Proc. of the Intemational Joint Conference on Neural Networks ( San Diego), vol. 2, 631–637, 1990Google Scholar
  5. [5]
    M.R. Choi, Implementation of Feedforward Artificial Neural Networks WithLeaming Using Standard CMOS Technology, PhD. Thesis, Michigan State University, 1991Google Scholar
  6. [6]
    J. Choi, S.H. Bang and B.J. Sheu, “A programmable Analog VLSI Neural Network Processor for Communication Receivers”, IEEE Transactions on Neural Networks, 484–495, 1993Google Scholar
  7. [7]
    W.C. Fang, B.J. Sheu, O.T.C. Chen and J. Choi, “A VLSI Neural Processor for Image Data Compression Using Self-Organization Networks”, IEEE tr. Neural Networks, vol. 3, 1992Google Scholar
  8. [8]
    F.J. Kub, K.K. Moon, I.A. Mack and F.M. Long, “Programmable Analog Vector-Matrix Multipliers”, IEEE Journal of Solid-State Circuits, 207–214, 1990Google Scholar
  9. [9]
    J.A. Lansner and T. Lehmann, “A Neuron-and Synapse Chip for Artificial Neural Networks”, In Proceeding of the European Solid-State Circuits Conference (Copenhagen), 213–216, 1992Google Scholar
  10. [10]
    J.A. Lansner and T. Lehmann, “An Analog CMOS Chip Set for Neural Networks with Arbitrary Topologies”, IEEE Transactions on Neural Networks, 441–445, 1993Google Scholar
  11. [11]
    J.C. Lee, B.J. Sheu, W.C. Fang and R Chelappa, “VLSI Neuroprocessors for Video Motion Detection”, IEEE Transactions on Neural Networks, 178–190, 1993Google Scholar
  12. [12]
    B. Linares-Barranco, E.S ánches-Sinencio, A Rodríguez-Vázquez and J.L. Huertas, “VLSI Implementation of a Transconductance mode Continuous BAM with On Chip Learning and Dynamic Analog Memory”, In Proceeding of the IEEE International Symposium on Circuits and Systems (Singapore), 1283–1286, 1991Google Scholar
  13. [13]
    B. Linares-Barranco, E. Sánches-Sinencio, A Rodríguez-Vázquez and J.L. Huertas, “A CMOS Analog Adaptive BAM with On-Chip Learning and Weight Refreshing”, IEEE Transactions on Neural Networks, 445–455, 1993Google Scholar
  14. [14]
    J.B. Lont and W. Guggenbühl, “Analog CMOS Implementation of a Multilayer Perceptron with Nonlinear Synapses”, IEEE Transactions on Neural Networks, 457–465, 1992Google Scholar
  15. [15]
    T.C. Lu, M.L. Chiang and J.B. Kuo, “A One-Transistor Synapse Circuit with an Analog LMS Adaptive Feedback for Neural Network VLSI”, In Proceedings of the IEEE International Symposium on Circuits and Systems (Singapore), 1303–1306, 1991Google Scholar
  16. [16]
    F.J. Pelayo, B. Pino, A Prieto, J. Ortega and F.J. Fernandez, “Analog Implementation of a Synaptic Matrix with Alterable Weight”, In Proceedings of the 2nd International Conference on Microelectronics for Neural Networks, Munich: Kyrill & Method Verlag, 261–267, 1991Google Scholar
  17. [17]
    F.M.A. Salam, M.R. Choi and Y. Wang, “Artificial Neural Nets in MOS Silicon”, In (Sethi & Jain, eds.) Artificial Neural Networks and Statistical Pattern Recognition, 243–270, 1991Google Scholar
  18. [18]
    S. Saryanarayana, Y.P. Tsividis and H.P. Graf, “A Reconfigurable VLSI Neural Network”, IEEE Journal of Solid-State Circuits, 67–81, 1992Google Scholar
  19. [19]
    B.J. Sheu, “VLSI Neurocomputing with Analog Programmable Chips and Digital Systolic Array Chips”, In Proceedings of the IEEE International Symposium on Circuits and Systems (Singapore), 1267–1270, 1991Google Scholar
  20. [20]
    L. Song, M.I. Elmasry and A. Vanelli, “Analog Neural Network Building Blocks Based on Current Mode Subthreshold Operation”, In Proceedings of the IEEE International Symposium on Circuits and Systems (Chicago), 2462–2465, 1993Google Scholar
  21. [21]
    Y. Wang, “AModular Analog CMOS LSI for Feedforward Neural Networks with On-Chip BEP Learning”, in Proceeding of the IEEE International Symposium on Circuits and Systems (Chicago), 2744 2747, 1993Google Scholar
  22. [22]
    R.J. Wiegerink, “A CMOS analog four-quadrant multiplier”, In Proceedings of the IEEE International Symposium on Circuits and Systems (Singapore), 2244–2247, 1991Google Scholar
  23. [23]
    H.C. Eaton and T.L. Oliver, “Learning Coefficient Dependence on Training Set Size”, Neural Networks, vol. 5, pp. 283–288, 1992CrossRefGoogle Scholar
  24. [24]
    S. Roy, “Factors influencing the choice of a learning rate for a backpropagation neural network”, in Proc. IEEE ICNN’94, pp. 503–507, Orlando, Fla, 1994Google Scholar

Copyright information

© Springer Science+Business Media New York 1995

Authors and Affiliations

  • Anne-Johan Annema
    • 1
  1. 1.MESA Research InstituteUniversity of TwenteNetherlands

Personalised recommendations