Convergence Analysis of Mean Shift

被引：0
|
作者：
Yamasaki, Ryoya [1 ]
Tanaka, Toshiyuki [2 ]
机构：
[1] Hitotsubashi Univ, Hitotsubashi Inst Adv Study, Tokyo 1868601, Japan
[2] Kyoto Univ, Grad Sch Informat, Dept Informat, Kyoto 6068501, Japan
来源：
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE | 2024年 / 46卷 / 10期
关键词：
Kernel; Convergence; Heuristic algorithms; Estimation; Clustering algorithms; Optimization; Bandwidth; Biweight kernel; convergence; convergence rate; mean shift; Lojasiewicz inequality; DESCENT METHODS; SUBANALYTIC FUNCTIONS; DENSITY-FUNCTION; ALGORITHM;
D O I：
10.1109/TPAMI.2024.3385920
中图分类号：
TP18 [人工智能理论];
学科分类号：
081104 ; 0812 ; 0835 ; 1405 ;
摘要：
The mean shift (MS) algorithm seeks a mode of the kernel density estimate (KDE). This study presents a convergence guarantee of the mode estimate sequence generated by the MS algorithm and an evaluation of the convergence rate, under fairly mild conditions, with the help of the argument concerning the Lojasiewicz inequality. Our findings extend existing ones covering analytic kernels and the Epanechnikov kernel. Those are significant in that they cover the biweight kernel, which is optimal among non-negative kernels in terms of the asymptotic statistical efficiency for the KDE-based mode estimation.
引用
收藏
页码：6688 / 6698
页数：11
相关论文
共 50 条
12345