Bayesian interpretation to generalize adaptive mean shift algorithm

Article type: Research Article

Authors: Yoon, Ji Won^*

Affiliations: Center for information security technology (CIST), Korea University, Republic of Korea

Correspondence: [*] Corresponding author. Ji Won Yoon, Center for information security technology (CIST), Korea University, Republic of Korea. Tel.: +82 2 3290 4886; E-mail: [email protected].

Abstract: The Adaptive Mean Shift (AMS) algorithm is a popular and simple non-parametric clustering approach based on Kernel Density Estimation. In this paper the AMS is reformulated in a Bayesian framework, which permits a natural generalization in several directions and is shown to improve performance. The Bayesian framework considers the AMS to be a method of obtaining a posterior mode. This allows the algorithm to be generalized with three components which are not considered in the conventional approach: node weights, a prior for a particular location, and a posterior distribution for the bandwidth. Practical methods of building the three different components are considered.

Keywords: Adaptive mean shift algorithm, kernel density estimation

DOI: 10.3233/IFS-162103

Journal: Journal of Intelligent & Fuzzy Systems, vol. 30, no. 6, pp. 3583-3592, 2016