Selecting the Minkowski Exponent for Intelligent K-Means with Feature Weighting

Renato Cordeiro De Amorim; Boris Mirkin

doi:10.1007/978-1-4939-0742-7_7

Selecting the Minkowski Exponent for Intelligent K-Means with Feature Weighting

Renato Cordeiro De Amorim, Boris Mirkin

Research output: Chapter in Book/Report/Conference proceeding › Chapter (peer-reviewed) › peer-review

5 Citations (Scopus)

Abstract

Recently, a three-stage version of K-Means has been introduced, at which not only clusters and their centers, but also feature weights are adjusted to minimize the summary p-th power of the Minkowski p-distance between entities and centroids of their clusters. The value of the Minkowski exponent p appears to be instrumental in the ability of the method to recover clusters hidden in data. This paper advances into the problem of finding the best p for a Minkowski metric-based version of K-Means, in each of the following two settings: semi-supervised and unsupervised. This paper presents experimental evidence that solutions found with the proposed approaches are sufficiently close to the optimum.

Original language	English
Title of host publication	Clusters, Orders, and Trees
Subtitle of host publication	Methods and Applications
Publisher	Springer
Pages	103-117
ISBN (Electronic)	978-1-4939-0742-7
ISBN (Print)	978-1-4939-0741-0
DOIs	https://doi.org/10.1007/978-1-4939-0742-7_7
Publication status	Published - May 2014

Publication series

Name	Springer Optimization and Its Applications
Volume	92

Access to Document

10.1007/978-1-4939-0742-7_7

Cite this

@inbook{20195da1512a4358b2b8efe1cc8a4990,

title = "Selecting the Minkowski Exponent for Intelligent K-Means with Feature Weighting",

abstract = "Recently, a three-stage version of K-Means has been introduced, at which not only clusters and their centers, but also feature weights are adjusted to minimize the summary p-th power of the Minkowski p-distance between entities and centroids of their clusters. The value of the Minkowski exponent p appears to be instrumental in the ability of the method to recover clusters hidden in data. This paper advances into the problem of finding the best p for a Minkowski metric-based version of K-Means, in each of the following two settings: semi-supervised and unsupervised. This paper presents experimental evidence that solutions found with the proposed approaches are sufficiently close to the optimum.",

author = "{Cordeiro De Amorim}, Renato and Boris Mirkin",

year = "2014",

month = may,

doi = "10.1007/978-1-4939-0742-7_7",

language = "English",

isbn = "978-1-4939-0741-0",

series = "Springer Optimization and Its Applications",

publisher = "Springer",

pages = "103--117",

booktitle = "Clusters, Orders, and Trees",

}

TY - CHAP

T1 - Selecting the Minkowski Exponent for Intelligent K-Means with Feature Weighting

AU - Cordeiro De Amorim, Renato

AU - Mirkin, Boris

PY - 2014/5

Y1 - 2014/5

N2 - Recently, a three-stage version of K-Means has been introduced, at which not only clusters and their centers, but also feature weights are adjusted to minimize the summary p-th power of the Minkowski p-distance between entities and centroids of their clusters. The value of the Minkowski exponent p appears to be instrumental in the ability of the method to recover clusters hidden in data. This paper advances into the problem of finding the best p for a Minkowski metric-based version of K-Means, in each of the following two settings: semi-supervised and unsupervised. This paper presents experimental evidence that solutions found with the proposed approaches are sufficiently close to the optimum.

AB - Recently, a three-stage version of K-Means has been introduced, at which not only clusters and their centers, but also feature weights are adjusted to minimize the summary p-th power of the Minkowski p-distance between entities and centroids of their clusters. The value of the Minkowski exponent p appears to be instrumental in the ability of the method to recover clusters hidden in data. This paper advances into the problem of finding the best p for a Minkowski metric-based version of K-Means, in each of the following two settings: semi-supervised and unsupervised. This paper presents experimental evidence that solutions found with the proposed approaches are sufficiently close to the optimum.

U2 - 10.1007/978-1-4939-0742-7_7

DO - 10.1007/978-1-4939-0742-7_7

M3 - Chapter (peer-reviewed)

SN - 978-1-4939-0741-0

T3 - Springer Optimization and Its Applications

SP - 103

EP - 117

BT - Clusters, Orders, and Trees

PB - Springer

ER -