Properties of the Nonparametric Decision Function with a Priori Information on Independence of Attributes of Classified Objects
A. V. Lapko1, V. A. Lapko2
1 Institute of Computational Modeling, Siberian Branch, Russian Academy of Sciences
2 Institute of Computational Modeling, Siberian Branch, Russian Academy of Sciences Reshetnev Siberian State Aerospace University
lapko@icm.krasn.ru, lapko@icm.krasn.ru
Keywords: pattern recognition, independent random variables, nonparametric estimate, asymptotic properties, a priori information
Pages: 112-119
Abstract
Asymptotic properties of a nonparametric estimate of the separating surface equation defined in the space of independent attributes of classified objects are studied. Based on this analysis, the importance of the influence of a priori information about independence of random variables on the approximation properties of the nonparametric decision function in a two-alternative problem of pattern recognition is established.
|
