Syntactic Neural Networks in VLSI

  • Simon Lucas
  • Bob Damper


Syntactic neural networks (SNNs) are an important new paradigm for neural computation. Their power stems from the existence of an underlying grammar which allows natural expression of time-varying patterns. In this paper, we are primarily concerned with the implementation of SNNs in VLSI. Their benefits for this purpose, compared to more common neural architectures, include sparse connections, fast learning and — for those nets based on a non-stochastic grammar — a simple and elegant digital realisation. We illustrate this with an example application; a dynamic signature verification system. Results are presented for a software simulation in which learning proceeds off-line. Hardware implementation is straightforward, requiring only a small number ofRAMs, shift-registers and adders.


Dynamic Time Warping VLSI Implementation Signature Verification Programmable Logic Array Weighted Euclidean Distance 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. Card, H.C. and Moore, W.R., “Implementation of plasticity in MOS synapses”, in Proceedings of IEE International Conference on Artificial Neural Networks, pp. 33–36, IEE, London, 1989.Google Scholar
  2. Fanty, M.A., “Context-free parsing with connectionist networks”, in Proceedings ofAIP Neural Networks for Computing Conference No. 151., pp. 140–145, Snowbird, UTAH, 1986.Google Scholar
  3. Knuth, D.E., “Semantics of context-free languages”, Journal of Mathematics and Systems Theory, 2, 127–146,1968.MathSciNetMATHCrossRefGoogle Scholar
  4. Lucas, S.M., Connectionist Architectures for Syntactic Pattern Recognition,PhD Thesis, Univeristy of Southampton, 1990a.Google Scholar
  5. Lucas, S.M., “Recent Progress on the VERISIGN Project”, Confidential Report to British Technology Group Limited, London, 1990b.Google Scholar
  6. Lucas, S.M. and Damper, R.I., “A new learning paradigm for neural networks”, Proceedings of IEE International Conference on Artificial Neural Networks, pp. 346–350, IEE, London, 1989a.Google Scholar
  7. Lucas, S.M. and Damper, R.I., “Using stochastic context-free grammars for modelling and recognising cursive script”, IEE Colloquium on Character Recognition,Digest No. 1989/109, IEE, London, 1989b.Google Scholar
  8. Lucas, S.M. and Damper, R.I., “Self-organising temporal networks”, Proceedings of IEEE Workshop on Genetic Algorithms,Neural Networks and Simulated Annealing Applied to Problems in Signal and Image Processing, IEEE, Glasgow, 1990a.Google Scholar
  9. Lucas, S.M. and Damper, R.I., “Signature verification with syntactic neural nets”, Proceedings of the IJCNN-90, Vol. 1, pp. 373–378, IEEE, San Diego, CA, 1990b.Google Scholar
  10. Lucas, S.M. and Damper, R.I., “Syntactic neural networks”, Connection Science, 2, pp. 199–225, 1990c.CrossRefGoogle Scholar
  11. Lucas, S.M. and Damper, R.I., “Text-phonetics conversion with syntactic neural networks”, to appear in Proceedings of ESCA Workshop on Speech Synthesis,Autrans, France, 1990d.Google Scholar
  12. Murray, A.F., “Pulse arithmetic in VLSI neural networks”, IEEE Micro, July, pp. 64–74, 1989.Google Scholar
  13. Parizeau, M. and Plamondon, R., “A comparative analysis of regional correlation, dynamic time warping and skeletal tree matching for signature verification”, IEEE Transactions on Pattern Analysis and Machine Intelligence, 12, pp. 710–717, 1990.CrossRefGoogle Scholar
  14. Tomlinson, M.S., Walker, D.J. and Sivilotti, M.A., “A digital neural network architecture for VLSP”, Proceedings of IJCNN-90, Vol. 2, pp. 545–550, IEEE, San Diego, CA, 1990.Google Scholar

Copyright information

© Springer Science+Business Media New York 1991

Authors and Affiliations

  • Simon Lucas
    • 1
  • Bob Damper
    • 1
  1. 1.Department of Electronics and Computer ScienceUniversity of SouthamptonUK

Personalised recommendations