文献类型：期刊文献

英文题名：Implementing the competitive selection of the Matthew effect: An adaptive gradient neural network approach and its applications

作者：Feng, Zhuowen[1];Yang, Liu[2];Lin, Cong[2];Han, Lingbo[2];Wang, Guancheng[2]

机构：[1]Guangdong Ocean Univ, Coll Literature & News Commun, Zhanjiang 524088, Peoples R China;[2]Guangdong Ocean Univ, Coll Elect & Informat Engn, Zhanjiang 524088, Peoples R China

年份：2026

卷号：743

外文期刊名：INFORMATION SCIENCES

收录：SCI-EXPANDED(收录号：WOS:001717230100001)、、EI(收录号：20261020236438)、WOS

基金：This research was supported in part by National Natural Science Foundation of China under Grant 62541603, in part by Zhanjiang Non-funded Science and Technology Research Program under Grant 2025B01031; in part by the program for scientific research startup funds of Guangdong Ocean University (Grant No. 060302112401 and 060302112201) ; in part by the Undergraduate Innovation Team Project of Guangdong Ocean University under Grant CXTD2024011 and JDTD2024003.

语种：英文

外文关键词：Matthew effect; Gradient neural network; k-winners-take-all (k-WTA); Adaptive coefficients

外文摘要：The Matthew Effect is pervasive across various domains, such as marketing, social networks, and multi-robot systems. One important aspect of this phenomenon is competitive selection, which can be formulated through a k-Winners-Take-All (k-WTA) mechanism. In this study, we provide a computational abstraction of this competitive selection mechanism using the k-WTA model and reformulate it as an equivalent quadratic programming problem and a nonlinear equation, thereby converting the task into a zero-finding problem. Against this backdrop, this study introduces a novel gradient-based neural framework featuring adaptive coefficients to implement the k-WTA model, achieving superior convergence rate and accuracy. Furthermore, theoretical analyses and empirical evidence confirm that our proposed model can effectively realize the k-WTA model and outperform alternative models. Finally, we illustrate the effects of k-WTA-based competitive selection in opinion evolution and robotic motion planning, demonstrating the flexibility of the proposed framework.