Scholarlink

<p>The minimum error entropy with fiducial points (MEEF) criterion has received increasing attention for its robustness in dealing with complex non-Gaussian noise environments. However, traditional MEEF often assumes that the center of the error distribution is fixed limiting performance under non-zero mean noise distribution. To address this issue, a variable-center MEEF (MEEF-VC) criterion is first defined in this paper, which can adaptively adjust the reference center of entropy estimation according to the characteristics of the error distribution to effectively capture non-zero-mean noise and dynamic error variations. Then, a novel robust diffusion MEEF-VC (DMEEF-VC) algorithm is developed by Adapt-Then-Combine strategy for distributed estimation over network. Moreover, the key parameters in DMEEF-VC (including bandwidth, center position and weighting factor) are optimized adaptively by probability density matching and entropy weight method. Simulation results show that the proposed DMEEF-VC exhibits robustness and accuracy enhancement in diffusion-based adaptive signal processing tasks compared with the traditional robust diffusion adaptive filtering algorithm.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Diffusion MEEF with Variable-Center Algorithm Optimized by Joint Probability Density Matching and Entropy Weight Method

  • 期刊论文

  • 发表日期：

    2025年9月24日

查看全文
  • Wentao Ma,
  • Wenbo Wu,
  • Peng Guo,
  • Badong Chen

摘要

The minimum error entropy with fiducial points (MEEF) criterion has received increasing attention for its robustness in dealing with complex non-Gaussian noise environments. However, traditional MEEF often assumes that the center of the error distribution is fixed limiting performance under non-zero mean noise distribution. To address this issue, a variable-center MEEF (MEEF-VC) criterion is first defined in this paper, which can adaptively adjust the reference center of entropy estimation according to the characteristics of the error distribution to effectively capture non-zero-mean noise and dynamic error variations. Then, a novel robust diffusion MEEF-VC (DMEEF-VC) algorithm is developed by Adapt-Then-Combine strategy for distributed estimation over network. Moreover, the key parameters in DMEEF-VC (including bandwidth, center position and weighting factor) are optimized adaptively by probability density matching and entropy weight method. Simulation results show that the proposed DMEEF-VC exhibits robustness and accuracy enhancement in diffusion-based adaptive signal processing tasks compared with the traditional robust diffusion adaptive filtering algorithm.