Skip to main content

2016 | OriginalPaper | Buchkapitel

Lower Bounds on Complexity of Shallow Perceptron Networks

verfasst von : Věra Kůrková

Erschienen in: Engineering Applications of Neural Networks

Verlag: Springer International Publishing

Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.

search-config
loading …

Abstract

Model complexity of shallow (one-hidden-layer) perceptron networks computing multivariable functions on finite domains is investigated. Lower bounds are derived on growth of the number of network units or sizes of output weights in terms of variations of functions to be computed. A concrete construction of a class of functions which cannot be computed by percetron networks with considerably smaller numbers of units and output weights than the sizes of the function’s domains is presented. In particular, functions on Boolean d-dimensional cubes are constructed which cannot be computed by shallow perceptron networks with numbers of hidden units and sizes of output weights depending on d polynomially. A subclass of these functions is described whose elements can be computed by two-hidden-layer networks with the number of units depending on d linearly.

Sie haben noch keine Lizenz? Dann Informieren Sie sich jetzt über unsere Produkte:

Springer Professional "Wirtschaft+Technik"

Online-Abonnement

Mit Springer Professional "Wirtschaft+Technik" erhalten Sie Zugriff auf:

  • über 102.000 Bücher
  • über 537 Zeitschriften

aus folgenden Fachgebieten:

  • Automobil + Motoren
  • Bauwesen + Immobilien
  • Business IT + Informatik
  • Elektrotechnik + Elektronik
  • Energie + Nachhaltigkeit
  • Finance + Banking
  • Management + Führung
  • Marketing + Vertrieb
  • Maschinenbau + Werkstoffe
  • Versicherung + Risiko

Jetzt Wissensvorsprung sichern!

Springer Professional "Technik"

Online-Abonnement

Mit Springer Professional "Technik" erhalten Sie Zugriff auf:

  • über 67.000 Bücher
  • über 390 Zeitschriften

aus folgenden Fachgebieten:

  • Automobil + Motoren
  • Bauwesen + Immobilien
  • Business IT + Informatik
  • Elektrotechnik + Elektronik
  • Energie + Nachhaltigkeit
  • Maschinenbau + Werkstoffe




 

Jetzt Wissensvorsprung sichern!

Springer Professional "Wirtschaft"

Online-Abonnement

Mit Springer Professional "Wirtschaft" erhalten Sie Zugriff auf:

  • über 67.000 Bücher
  • über 340 Zeitschriften

aus folgenden Fachgebieten:

  • Bauwesen + Immobilien
  • Business IT + Informatik
  • Finance + Banking
  • Management + Führung
  • Marketing + Vertrieb
  • Versicherung + Risiko




Jetzt Wissensvorsprung sichern!

Literatur
1.
Zurück zum Zitat Fine, T.L.: Feedforward Neural Network Methodology. Springer, Heidelberg (1999)MATH Fine, T.L.: Feedforward Neural Network Methodology. Springer, Heidelberg (1999)MATH
2.
Zurück zum Zitat Kecman, V.: Learning and Soft Computing. MIT Press, Cambridge (2001)MATH Kecman, V.: Learning and Soft Computing. MIT Press, Cambridge (2001)MATH
3.
Zurück zum Zitat LeCun, Y., Bottou, L., Bengio, Y., Haffner, P.: Gradient-based learning applied to document recognition. Proc. IEEE 86, 2278–2324 (1998)CrossRef LeCun, Y., Bottou, L., Bengio, Y., Haffner, P.: Gradient-based learning applied to document recognition. Proc. IEEE 86, 2278–2324 (1998)CrossRef
4.
5.
Zurück zum Zitat Bengio, Y.: Learning deep architectures for AI. Found. Trends Mach. Learn. 2, 1–127 (2009)CrossRefMATH Bengio, Y.: Learning deep architectures for AI. Found. Trends Mach. Learn. 2, 1–127 (2009)CrossRefMATH
6.
Zurück zum Zitat LeCunn, Y., Bengio, Y., Hinton, G.: Deep learning. Nature 521, 436–444 (2015)CrossRef LeCunn, Y., Bengio, Y., Hinton, G.: Deep learning. Nature 521, 436–444 (2015)CrossRef
7.
Zurück zum Zitat Ba, L.J., Caruana, R.: Do deep networks really need to be deep?. In: Ghahrani, Z., et al. (eds.) Advances in Neural Information Processing Systems, vol. 27, pp. 1–9 (2014) Ba, L.J., Caruana, R.: Do deep networks really need to be deep?. In: Ghahrani, Z., et al. (eds.) Advances in Neural Information Processing Systems, vol. 27, pp. 1–9 (2014)
8.
Zurück zum Zitat Kainen, P.C., Kůrková, V., Sanguineti, M.: Dependence of computational models on input dimension: tractability of approximation and optimization tasks. IEEE Trans. Inf. Theory 58, 1203–1214 (2012)MathSciNetCrossRef Kainen, P.C., Kůrková, V., Sanguineti, M.: Dependence of computational models on input dimension: tractability of approximation and optimization tasks. IEEE Trans. Inf. Theory 58, 1203–1214 (2012)MathSciNetCrossRef
9.
Zurück zum Zitat Maiorov, V., Pinkus, A.: Lower bounds for approximation by MLP neural networks. Neurocomputing 25, 81–91 (1999)CrossRefMATH Maiorov, V., Pinkus, A.: Lower bounds for approximation by MLP neural networks. Neurocomputing 25, 81–91 (1999)CrossRefMATH
10.
Zurück zum Zitat Bianchini, M., Scarselli, F.: On the complexity of neural network classifiers: a comparison between shallow and deep architectures. IEEE Trans. Neural Netw. Learn. Syst. 25(8), 1553–1565 (2014)CrossRef Bianchini, M., Scarselli, F.: On the complexity of neural network classifiers: a comparison between shallow and deep architectures. IEEE Trans. Neural Netw. Learn. Syst. 25(8), 1553–1565 (2014)CrossRef
11.
Zurück zum Zitat Bengio, Y., Delalleau, O., Roux, N.L.: The curse of highly variable functions for local kernel machines. In: Advances in Neural Information Processing Systems, vol. 18, pp. 107–114. MIT Press (2006) Bengio, Y., Delalleau, O., Roux, N.L.: The curse of highly variable functions for local kernel machines. In: Advances in Neural Information Processing Systems, vol. 18, pp. 107–114. MIT Press (2006)
12.
Zurück zum Zitat Kůrková, V., Sanguineti, M.: Model complexities of shallow networks representing highly varying functions. Neurocomputing 171, 598–604 (2016)CrossRef Kůrková, V., Sanguineti, M.: Model complexities of shallow networks representing highly varying functions. Neurocomputing 171, 598–604 (2016)CrossRef
13.
Zurück zum Zitat Ito, Y.: Finite mapping by neural networks and truth functions. Math. Sci. 17, 69–77 (1992)MathSciNetMATH Ito, Y.: Finite mapping by neural networks and truth functions. Math. Sci. 17, 69–77 (1992)MathSciNetMATH
14.
Zurück zum Zitat Barron, A.R.: Neural net approximation. In: Narendra, K. (ed.) Proceedings of the 7th Yale Workshop on Adaptive and Learning Systems, pp. 69–72. Yale University Press (1992) Barron, A.R.: Neural net approximation. In: Narendra, K. (ed.) Proceedings of the 7th Yale Workshop on Adaptive and Learning Systems, pp. 69–72. Yale University Press (1992)
15.
Zurück zum Zitat Kůrková, V.: Dimension-independent rates of approximation by neural networks. In: Warwick, K., Kárný, M. (eds.) Computer-Intensive Methods in Control and Signal Processing: The Curse of Dimensionality, pp. 261–270. Birkhäuser, Boston (1997)CrossRef Kůrková, V.: Dimension-independent rates of approximation by neural networks. In: Warwick, K., Kárný, M. (eds.) Computer-Intensive Methods in Control and Signal Processing: The Curse of Dimensionality, pp. 261–270. Birkhäuser, Boston (1997)CrossRef
16.
Zurück zum Zitat Kůrková, V., Sanguineti, M.: Comparison of worst-case errors in linear and neural network approximation. IEEE Trans. Inf. Theory 48, 264–275 (2002)MathSciNetCrossRefMATH Kůrková, V., Sanguineti, M.: Comparison of worst-case errors in linear and neural network approximation. IEEE Trans. Inf. Theory 48, 264–275 (2002)MathSciNetCrossRefMATH
17.
Zurück zum Zitat Kainen, P.C., Kůrková, V., Vogt, A.: A Sobolev-type upper bound for rates of approximation by linear combinations of heaviside plane waves. J. Approximation Theory 147, 1–10 (2007)MathSciNetCrossRefMATH Kainen, P.C., Kůrková, V., Vogt, A.: A Sobolev-type upper bound for rates of approximation by linear combinations of heaviside plane waves. J. Approximation Theory 147, 1–10 (2007)MathSciNetCrossRefMATH
18.
Zurück zum Zitat Kůrková, V.: Minimization of error functionals over perceptron networks. Neural Comput. 20, 250–270 (2008)MathSciNetMATH Kůrková, V.: Minimization of error functionals over perceptron networks. Neural Comput. 20, 250–270 (2008)MathSciNetMATH
19.
Zurück zum Zitat Kůrková, V.: Complexity estimates based on integral transforms induced by computational units. Neural Netw. 33, 160–167 (2012)CrossRefMATH Kůrková, V.: Complexity estimates based on integral transforms induced by computational units. Neural Netw. 33, 160–167 (2012)CrossRefMATH
20.
Zurück zum Zitat Gnecco, G., Sanguineti, M.: On a variational norm tailored to variable-basis approximation schemes. IEEE Trans. Inf. Theory 57, 549–558 (2011)MathSciNetCrossRef Gnecco, G., Sanguineti, M.: On a variational norm tailored to variable-basis approximation schemes. IEEE Trans. Inf. Theory 57, 549–558 (2011)MathSciNetCrossRef
21.
Zurück zum Zitat Kůrková, V., Savický, P., Hlaváčková, K.: Representations and rates of approximation of real-valued Boolean functions by neural networks. Neural Netw. 11, 651–659 (1998)CrossRef Kůrková, V., Savický, P., Hlaváčková, K.: Representations and rates of approximation of real-valued Boolean functions by neural networks. Neural Netw. 11, 651–659 (1998)CrossRef
22.
Zurück zum Zitat Cover, T.: Geometrical and statistical properties of systems of linear inequalities with applications in pattern recognition. IEEE Trans. Electron. Comput. 14, 326–334 (1965)CrossRefMATH Cover, T.: Geometrical and statistical properties of systems of linear inequalities with applications in pattern recognition. IEEE Trans. Electron. Comput. 14, 326–334 (1965)CrossRefMATH
23.
Zurück zum Zitat Erdös, P., Spencer, J.H.: Probabilistic Methods in Combinatorics. Academic Press, New York (1974)MATH Erdös, P., Spencer, J.H.: Probabilistic Methods in Combinatorics. Academic Press, New York (1974)MATH
25.
Zurück zum Zitat MacWilliams, F., Sloane, N.J.A.: The Theory of Error-Correcting Codes. North-Holland, Asterdam (1977)MATH MacWilliams, F., Sloane, N.J.A.: The Theory of Error-Correcting Codes. North-Holland, Asterdam (1977)MATH
Metadaten
Titel
Lower Bounds on Complexity of Shallow Perceptron Networks
verfasst von
Věra Kůrková
Copyright-Jahr
2016
DOI
https://doi.org/10.1007/978-3-319-44188-7_21

Premium Partner