Publisher: IGI Global_journal
E-ISSN: 1947-9336|8|1|58-92
ISSN: 1947-9328
Source: International Journal of Operations Research and Information Systems (IJORIS), Vol.8, Iss.1, 2017-01, pp. : 58-92
Disclaimer: Any content in publications that violate the sovereignty, the constitution or regulations of the PRC is not accepted or approved by CNPIEC.
Abstract
In this work, the authors propose a novel method called online variable kernel estimation of the probability density function (pdf). This new online estimator combines the characteristics and properties of two estimators namely nearest neighbors estimator and the Parzen-Rosenblatt estimator. Their approach allows a compact online adaptation of the estimated probability density function from the new arrival data. The performance of the online variable kernel estimator (OVKE) depends on the choice of the bandwidth. The authors present in this article a new technique for determining the optimal smoothing parameter of OVKE based on the maximum entropy principle (MEP). The robustness and performance of the proposed approach are demonstrated by examples of online estimation of real and simulated data distributions.