The problem of reconstructing dependencies from empirical data became very important in a very large range of applications. Procedures used to solve this problem are known as “Methods of Machine Learning” [1,3]. These procedures include methods of regression reconstruction, inverse problems of mathematical physics and statistics, machine learning in pattern recognition (for visual and abstract patterns represented by sets of features) and many others. Many web network control problems also belong to this field. The task is to reconstruct the dependency between input and output data as precisely as possible using empirical data obtained from experiments or statistical observations.


  1. 1.
    Vapnik, V.N., Chervonenkis, A.Y.: Theory of Pattern Recognition (in Russian). Moscow, Nauka (1974)Google Scholar
  2. 2.
    MacKay, J.D.C.: Bayesian interpolation. Neural Computation 4(3), 415–447 (1992)CrossRefzbMATHGoogle Scholar
  3. 3.
    Vapnik, V.N.: The Nature of Statistical Learning Theory. Springer, Heidelberg (2000)CrossRefzbMATHGoogle Scholar
  4. 4.
    Chervonenkis, A.Y.: A combined Bayes - Maximum Likelihood method for regression. In: Riccia, G.D., Lenz, H.-J., Kruse, R. (eds.) Data Fusion and Perception, Springer, Wien, New-York (2001)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • Alexei Ya. Chervonenkis
    • 1
  1. 1.Institute of Control SciencesMoscowRussia

Personalised recommendations