A New Algorithm for Learning Mahalanobis Discriminant Functions by a Neural Network

  • Yoshifusa Ito
  • Hiroyuki Izumi
  • Cidambi Srinivasan
Part of the Lecture Notes in Computer Science book series (LNCS, volume 7063)


It is well known that a neural network can learn Bayesian discriminant functions. In the two-category normal-distribution case, a shift by a constant of the logit transform of the network output approximates a corresponding Mahalanobis discriminant function [7]. In [10], we have proposed an algorithm for estimating the constant, but it requires the network to be trained twice, in one of which the teacher signals must be shifted by the mean vectors. In this paper, we propose a more efficient algorithm for estimating the constant with which the network is trained only once.


Neural Network Discriminant Function Mahalanobis Distance Output Unit Network Output 
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. 1.
    Duda, R.O., Hart, P.E.: Pattern classification and scene analysis. John Wiley & Sons, New York (1973)zbMATHGoogle Scholar
  2. 2.
    Funahashi, K.: Multilayer neural networks and Bayes decision theory. Neural Networks 11, 209–213 (1998)CrossRefGoogle Scholar
  3. 3.
    Ito, Y.: Simultaneous approximations of polynomials and derivatives and their applications to neural networks. Neural Computation 20, 2757–2791 (2008)MathSciNetCrossRefzbMATHGoogle Scholar
  4. 4.
    Ito, Y., Srinivasan, C.: Multicategory Bayesian Decision Using a Three-Layer Neural Network. In: Kaynak, O., Alpaydın, E., Oja, E., Xu, L. (eds.) ICANN 2003 and ICONIP 2003. LNCS, vol. 2714, pp. 253–261. Springer, Heidelberg (2003)CrossRefGoogle Scholar
  5. 5.
    Ito, Y., Srinivasan, C.: Bayesian decision theory on three-layer neural networks. Neurocomputing 63, 209–228 (2005)CrossRefGoogle Scholar
  6. 6.
    Ito, Y., Srinivasan, C., Izumi, H.: Bayesian Learning of Neural Networks Adapted to changes of Prior Probabilities. In: Duch, W., Kacprzyk, J., Oja, E., Zadrożny, S. (eds.) ICANN 2005. LNCS, vol. 3697, pp. 253–259. Springer, Heidelberg (2005)Google Scholar
  7. 7.
    Ito, Y., Srinivasan, C., Izumi, H.: Discriminant Analysis by a Neural Network with Mahalanobis Distance. In: Kollias, S.D., Stafylopatis, A., Duch, W., Oja, E. (eds.) ICANN 2006. LNCS, vol. 4132, pp. 350–360. Springer, Heidelberg (2006)CrossRefGoogle Scholar
  8. 8.
    Ito, Y., Srinivasan, C., Izumi, H.: Learning of Bayesian Discriminant Functions by a Neural Network. In: Ishikawa, M., Doya, K., Miyamoto, H., Yamakawa, T. (eds.) ICONIP 2007, Part I. LNCS, vol. 4984, pp. 238–247. Springer, Heidelberg (2008)CrossRefGoogle Scholar
  9. 9.
    Ito, Y., Srinivasan, C., Izumi, H.: Multi-Category Bayesian Decision by Neural Networks. In: Kůrková, V., Neruda, R., Koutník, J. (eds.) ICANN 2008, Part I. LNCS, vol. 5163, pp. 21–30. Springer, Heidelberg (2008)CrossRefGoogle Scholar
  10. 10.
    Ito, Y., Srinivasan, C., Izumi, H.: Learning of Mahalanobis Discriminant Functions by a Neural Network. In: Leung, C.S., Lee, M., Chan, J.H. (eds.) ICONIP 2009. LNCS, vol. 5863, pp. 417–424. Springer, Heidelberg (2009)CrossRefGoogle Scholar
  11. 11.
    Richard, M.D., Lipmann, R.P.: Neural network classifiers estimate Bayesian a posteriori probabilities. Neural Computation 3, 461–483 (1991)CrossRefGoogle Scholar
  12. 12.
    Ruck, M.D., Rogers, S., Kabrisky, M., Oxley, H., Sutter, B.: The multilayer perceptron as approximator to a Bayes optimal discriminant function. IEEE Transactions on Neural Networks 1, 296–298 (1990)CrossRefGoogle Scholar
  13. 13.
    White, H.: Learning in artificial neural networks: A statistical perspective. Neural Computation 1, 425–464 (1989)CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • Yoshifusa Ito
    • 1
  • Hiroyuki Izumi
    • 2
  • Cidambi Srinivasan
    • 3
  1. 1.School of MedicineAichi Medical UniversityNagakuteJapan
  2. 2.Department of Policy ScienceAichi-Gakuin UniversityNisshinJapan
  3. 3.Department of StatisticsUniversity of KentuckyLexingtonUSA

Personalised recommendations