Abstract
We propose a new approach for solving the classification problem, which is based on the using \(\epsilon \)-nets theory. It is showed that for separating two sets one can use their \(\epsilon \)-nets, which considerably reduce the complexity of the separating algorithm for large sets’ sizes. The necessary and sufficient conditions of separable \(\epsilon \)-nets of two sets are proved. The algorithm of building separable \(\epsilon \)-nets is proposed. The \(\epsilon \)-nets, constructed according to this algorithm, have size \(O(1/\varepsilon )\), which does not depended on the size of set. The set of possible values of \(\epsilon \) for \(\epsilon \)-nets of both sets is considered. The properties of this set and the theorem of its convergence are proved. The proposed algorithm of solving the classification problem using the \(\epsilon \)-nets has the same computational complexity as Support Vector Machine \(O(n\ln n)\) and its accuracy is comparable with SVM results.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Haussler, D., Welzl, E.: Epsilon-nets and simplex range queries. Discrete Comput. Geom. 2, 127–151 (1987)
Ivanchuk, M.A., Malyk, I.V.: Using-nets for linear separation of two sets in a Euclidean space \(R^{d}\). Cybern. Syst. Anal. 51(6), 965–968 (2015)
Tucker, H.G.: Generalization of the Glivenko-Cantelli theorem. Ann. Math. Stat. 30(3), 828–830 (1959)
Dvoretzky, A., Kiefer, J., Wolfowitz, J.: Asymptotic minimax character of the sample distribution function and of the classical multinomial estimator. Ann. Math. Stat. 27(3), 642–669 (1956)
Benedetto, J.J., Czaja, W.: Integration and Modern Analysis. Birkhäuser Advanced Texts Basler Lehrbücher. Springer, Heidelberg (2009). pp. 361–364
Durrett, R.: Probability: Theory and Examples, 4.1st edn. Cambridge University Press, Cambridge (2013). 386 p.
Ivanchuk, M.A., Malyk, I.V.: Separation of convex hulls as a way for modeling of systems of prediction of complications in patients. J. Autom. Inf. Sci. 47(4), 78–84 (2015)
Christopher, J.C.: Tutorial on support vector machines for pattern recognition. Data Min. Knowl. Disc. 2(2), 121–167 (1998)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2017 Springer International Publishing AG
About this paper
Cite this paper
Ivanchuk, M.A., Malyk, I.V. (2017). Using \(\epsilon \)-nets for Solving the Classification Problem. In: Dimov, I., Faragó, I., Vulkov, L. (eds) Numerical Analysis and Its Applications. NAA 2016. Lecture Notes in Computer Science(), vol 10187. Springer, Cham. https://doi.org/10.1007/978-3-319-57099-0_40
Download citation
DOI: https://doi.org/10.1007/978-3-319-57099-0_40
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-57098-3
Online ISBN: 978-3-319-57099-0
eBook Packages: Computer ScienceComputer Science (R0)