Abstract
The work is devoted to a problem of statistical robustness of deciding functions, or risk estimation. By risk we mean some measure of decision function prediction quality, for example, an error probability. For the case of discrete “independent” variable the dependence of average risk on empirical risk for the “worst” distribution (“strategies of nature”) is obtained. The result gives exact value of empirical risk bias that allows evaluating an accuracy of Vapnik-Chervonenkis risk estimations. To find a distribution providing maximum of empirical risk bias one need to solve an optimization problem on function space. The problem being very complicate in general case appears to be solvable when the “independent” feature is a space of isolated points. The space has low practical use but it allows scaling well-known estimations by Vapnik and Chervonenkis.
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
Vapnik, V.N., Chervonenkis, A. Ja.: Theory of pattern recognition. “Nauka”. Moscow (1974) 415p. (in Russian)
Hyghes, G. F.: On the mean accuracy of statistical pattern recognizers. IEEE Trans. Inform. Theory. V. IT-14, N 1. (1968) 55–63
Lbov, G.S., Startseva, N.G.: Logical deciding functions and questions of statistical stability of decisions. Institute of mathematics. Novosibirsk (1999) 211p. (in Russian).
Nedel’ko, V.M.: An Asymptotic Estimate of the Quality of a Decision Function Based on Empirical Risk for the Case of a Discrete Variable. Pattern Recognition and Image Analysis. Vol. 11, No. 1 (2001) 69–72
Berikov, V.B.: On stability of recognition algorithms in discrete statement. Artificial intelligence. Ukraine. N. 2. (2000) 5–8 (in Russian)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2003 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Nedel’ko, V.M. (2003). Estimating a Quality of Decision Function by Empirical Risk. In: Perner, P., Rosenfeld, A. (eds) Machine Learning and Data Mining in Pattern Recognition. MLDM 2003. Lecture Notes in Computer Science, vol 2734. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-45065-3_16
Download citation
DOI: https://doi.org/10.1007/3-540-45065-3_16
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-40504-7
Online ISBN: 978-3-540-45065-8
eBook Packages: Springer Book Archive