Kernel Least Mean Square with Adaptive Kernel Size
This work addresses a specific bottleneck in kernel adaptive filters for researchers and practitioners, offering an incremental improvement over manual or pre-estimated kernel size methods.
The paper tackled the problem of selecting the kernel size in kernel least mean square algorithms by developing an online optimization technique, resulting in improved performance confirmed through simulations on static function estimation and chaotic time series prediction.
Kernel adaptive filters (KAF) are a class of powerful nonlinear filters developed in Reproducing Kernel Hilbert Space (RKHS). The Gaussian kernel is usually the default kernel in KAF algorithms, but selecting the proper kernel size (bandwidth) is still an open important issue especially for learning with small sample sizes. In previous research, the kernel size was set manually or estimated in advance by Silvermans rule based on the sample distribution. This study aims to develop an online technique for optimizing the kernel size of the kernel least mean square (KLMS) algorithm. A sequential optimization strategy is proposed, and a new algorithm is developed, in which the filter weights and the kernel size are both sequentially updated by stochastic gradient algorithms that minimize the mean square error (MSE). Theoretical results on convergence are also presented. The excellent performance of the new algorithm is confirmed by simulations on static function estimation and short term chaotic time series prediction.