Current browse context:
stat.ML
Change to browse by:
References & Citations
Statistics > Machine Learning
Title: A Sub-Quadratic Exact Medoid Algorithm
(Submitted on 23 May 2016 (v1), revised 30 May 2016 (this version, v2), latest version 12 Apr 2017 (v4))
Abstract: We present a new algorithm, trimed, for obtaining the medoid of a set, that is the element of the set which minimises the mean distance to all other elements. The algorithm is shown to have, under weak assumptions, complexity O(N^(3/2)) in R^d where N is the set size, making it the first sub-quadratic exact medoid algorithm for d>1. Experiments show that it performs very well on spatial network data, frequently requiring two orders of magnitude fewer distances than state-of-the-art approximate algorithms. We show how trimed can be used as a component in an accelerated K-medoids algorithm, and how it can be relaxed to obtain further computational gains with an only minor loss in quality.
Submission history
From: James Newling [view email][v1] Mon, 23 May 2016 09:24:59 GMT (482kb,D)
[v2] Mon, 30 May 2016 07:44:29 GMT (482kb,D)
[v3] Tue, 11 Apr 2017 07:55:33 GMT (903kb,D)
[v4] Wed, 12 Apr 2017 18:25:34 GMT (904kb,D)
Link back to: arXiv, form interface, contact.