A Complete Investigation of Using Weighted Kernel Regression for The Case of Small Sample Problem With Noise
Zuwairie Ibrahim, Mohd Ibrahim Shapiai, Siti Nurzulaikha Satiman, Mohd Saberi Mohamad, Nurul Wahidah Arshad · ARPN Journal of Engineering and Applied Sciences · 2015
Weighted kernel regression (WKR) is a kernel-based regression approach for small sample problems. Previously, for the case of small sample problems with noise, we have done preliminary studies which investigated different learning techniques and different learning functions, separately. In this paper, a complete investigation of using WKR for the case of noisy and small training samples is presented. Analysis and discussion are provided in detail.