Quantized kernel maximum correntropy and its mean square convergence analysis
Quantized kernel maximum correntropy and its mean square convergence analysis
复制标题
量化核最大相关熵及其均方收敛性分析
DOI：
10.1016/j.dsp.2017.01.010
复制
发表时间：
2017-04-01
影响因子：
2.9
通讯作者：
Tan, Hongtao
中科院分区：
文献类型：
--
作者：
Wang, Shiyuan;Zheng, Yunfei;Tan, Hongtao
在线向量量化（VQ）方法已成功应用于核自适应滤波器（KAFs），以抑制其线性增长的径向基函数（RBF）网络，从而产生了一系列量化核自适应滤波器（QKAFs）。然而，现有的大多数QKAFs基于均方误差（MSE）准则，这对于非高斯信号实际上并非一个好的选择。在本文中，开发了一种新的量化核自适应滤波器，称为量化核最大相关熵（QKMC），它对大的异常值或脉冲噪声具有鲁棒性。对QKMC进行了均方收敛分析，从而得到了保证收敛的充分条件。还证明了QKMC的滤波精度高于具有代表性的量化核最小均方（QKLMS）。此外，为了充分利用隐藏在输入和输出空间中的信息，我们进一步提出了一种基于双边梯度的改进型QKMC（QKMC - BG）。为了限制QKMC - BG的最终网络规模，还开发了具有固定预算的QXMC - BG（QKMC - BG - FB）。给出了在高斯和非高斯噪声情况下的仿真结果，以验证所提出的QKMC、QKMC - BG和QKMC - BG - FB。（C）2017爱思唯尔公司。保留所有权利。
Online vector quantization (VQ) method has been successfully applied to the kernel adaptive filters (KAFs) for curbing their linearly growing radial basis function (RBF) network, thereby generating a family of quantized KAFs (QKAFs). However, the most existing QKAFs are based on the mean square error (MSE) criterion, which is actually not a good choice for non-Gaussian signals. In this paper, a new quantized kernel adaptive filter called quantized kernel maximum correntropy (QKMC) is developed, which is robust to large outliers or impulsive noises. The mean square convergence analysis for QKMC is conducted, and a sufficient condition for guaranteeing convergence is therefore obtained. The filtering accuracy of QKMC is also proved to be higher than that of the representative quantized kernel least mean square (QKLMS). In addition, to make full use of the information hidden in the input and the output spaces, we further propose a modified QKMC based on bilateral gradient (QKMC-BG). To limit the final network size of QKMC-BG, the QXMC-BG with fixed budget (QKMC-BG-FB) is also developed. Simulation results under the cases of Gaussian and non-Gaussian noises are presented to validate the proposed QKMC, QKMC-BG and QKMC-BG-FB. (C) 2017 Elsevier Inc. All rights reserved.