Abstract
The paper presents a theoretical proof revealing an intrinsic limitation of digital VLSI technology: its inability to cope with highly connected structures (e.g. neural networks). We are in fact able to prove that efficient digital VLSI implementations (known as VLSI-optimal when minimising the AT 2 complexity measure — A being the area of the chip, and T the delay for propagating the inputs to the outputs) of neural networks are achieved for small-constant fan-in gates. This result builds on quite recent ones dealing with a very close estimate of the area of neural networks when implemented by threshold gates, but it is also valid for classical Boolean gates. Limitations and open questions are presented in the conclusions.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Y.S. Abu-Mostafa, Connectivity Versus Entropy, in: Neural Information Processing Systems (Proc. NIPS*87, Denver, Colorado), ed. D.Z. Anderson, American Institute of Physics, New York, (1988) pp1–8.
V. Beiu, J.A. Peperstraete, J. Vandewalle and R. Lauwereins, Efficient Decomposition of COMPARISON and Its Applications, in: ESANN’93 (Proc. European Symposium on Artificial Neural Networks ‘93, Brussels, Belgium), ed. M. Verleysen, Dfacto, Brussels, (1993) pp45–50.
V. Beiu, J.A. Peperstraete, J. Vandewalle and R. Lauwereins, COMPARISON and Threshold Gate Decomposition, in: MicroNeuro’ 93 (Proc. International Conference on Microelectronics for Neural Networks’ 93, Edinburgh, UK), eds. D.J. Myers and A.F. Murray, UnivEd Tech. Ltd., Edinburgh, (1993) pp83–90.
V. Beiu, J.A. Peperstraete, J. Vandewalle and R. Lauwereins, Learning from Examples and VLSI Implementation of Neural Networks, in: Cybernetics and System Research’ 94 (Proc. European Meeting on Cybernetics and System Research’ 94, Vienna, Austria), ed. R. Trappl, World Scientific Publishing, Singapore, (1994) pp1767–1774.
V. Beiu, J.A. Peperstraete, J. Vandewalle and R. Lauwereins, Area-Time Performances of Some Neural Computations, in: SPRANN’ 94 (Proc. International Symposium on Signal Processing, Robotics and Neural Networks’ 94, Lille, France), eds. P. Borne, T. Fukuda and S.G. Tzafestas, GERF EC, Lille, (1994) pp664–668.
V. Beiu and J.G. Taylor, VLSI Optimal Neural Network Learning Algorithm, in: Artificial Neural Nets and Genetic Algorithms (Proc. Int. Conf., Ales, France), eds. D.W. Pearson, N.C. Steele and R.F. Albrecht, Springer-Verlag, Vienna, (1995) pp61–64.
V. Beiu and J.G. Taylor, Area-Efficient Constructive Learning Algorithms, in Proc. CSCS-10 (10th International Conference on Control System and Computer Science, Bucharest, România), ed. I. Dumitrache, PUBucharest, Bucharest, (1995), pp293–310.
J. Bruck and J. Goodman, On the Power of Neural Networks for Solving Hard Problems, in: Neural Information Processing Systems (Proc. NIPS*87, Denver, Colorado), ed. D.Z. Anderson, American Institute of Physics, New York, (1988) pp137–143.
D. Hammerstrom, The Connectivity Analysis of Simple Associations-or-How Many Connections Do You Need, in: Neural Information Processing Systems (Proc. NIPS*87, November, Denver, Colorado), ed. D.Z. Anderson, American Institute of Physics, New York, (1988) pp338–347.
H. Klaggers and M. Soegtrop, Limited Fan-In Random Wired Cascade-Correlation, in: MicroNeuro’93 (Proc. International Conference on Microelectronics for Neural Networks, Edinburgh’ 93, UK), eds. D.J.6Myers and A.F. Murray, UnivEd Tech. Ltd., Edinburgh, (1993) pp79–82.
A.V. Krishnamoorthy, R. Paturi, M. Blume, G.D. Linden, L.H. Linden and S.C. Esener, Hardware Tradeoffs for Boolean Concept Learning, in WCNN’94 (Proc. World Conference on Neural Networks’ 94, San Diego, USA), Lawrence Erlbaum and INNS Press, Hillsdale, (1994) Vol. 1, pp551–559.
D.S. Phatak and I. Koren, Connectivity and Performance Tredeoffs in the Cascade-Correlation Learning Architecture, IEEE Trans. on Neural Networks Vol. 5(6) (1994), pp930–935.
N.P. Red’kin, Synthesis of Threshold Circuits for certain Classes of Boolean Functions, Kibernetica Vol. 5 (1970), pp6–9. Translated in Cybernetics Vol. 6(5) (1973), pp540-544.
K.-Y. Siu, V. Roychowdhury and T. Kailath, Depth-Size Tradeoffs for Neural Computations, IEEE Trans. on Computers Vol. 40(12) (1991), pp1402–1412.
R.C. Williamson, ε-Entropy and the Complexity of Feedforward Neural Networks, in: Neural Information Processing Systems (Proc. NIPS*90, Denver, Colorado), eds. R.P. Lippmann, J.E. Moody and D.S. Touretzky, Morgan Kaufmann, San Mateo, (1991), pp946–952.
B.-T. Zhang and H. Mühlenbein, Genetic Programming of Minimal Neural Networks Using Occam’s Razor, Technical Report: Arbeitspapiere der GMD 734, Schloß Birlinghoven, Sankt Augustin, Germany (1993).
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1997 Springer Science+Business Media New York
About this chapter
Cite this chapter
Beiu, V. (1997). Constant Fan-in Digital Neural Networks are VLSI-Optimal. In: Ellacott, S.W., Mason, J.C., Anderson, I.J. (eds) Mathematics of Neural Networks. Operations Research/Computer Science Interfaces Series, vol 8. Springer, Boston, MA. https://doi.org/10.1007/978-1-4615-6099-9_12
Download citation
DOI: https://doi.org/10.1007/978-1-4615-6099-9_12
Publisher Name: Springer, Boston, MA
Print ISBN: 978-1-4613-7794-8
Online ISBN: 978-1-4615-6099-9
eBook Packages: Springer Book Archive