Author(s)

Yiwen Xu, Jing Chen, Zhongwen Zhuang, Liqun Lin

Corresponding Author

Yiwen Xu

Abstract

Kernel Density Estimation (KDE) is an effective method to estimate the probability density of random data, and is often used in the data processing in applications. In the long-term scenario, it is necessary to update KDE model to ensure its accuracy. However, the most traditional update method in KDE, random resampling, has the weakness of ignoring the contributions of the KDE models used before. Thus, it has negative influence on its accuracy and computation complexity. To solve these problems, we present a new update method based on human forgetting mechanism in this paper. Furthermore, we conduct comparative experiments based on the scenario of Internet of Vehicles (IoV) to verify the performance of the proposed method. The experimental results show that our proposed method is superior to the random resampling.

Keywords

Kernel Density Estimation (KDE), Probability Density Function (PDF), human forgetting mechanism, Internet of Vehicles (IoV)