Skoči na glavni sadržaj

KoG, Vol. 20 No. 20, 2016.

Stručni rad

k-means Algorithm

Ivančica Mirošević orcid id orcid.org/0000-0002-6356-1943 ; Fakultet elektrotehnike, strojarstva i brodogradnje Sveučilišta u Splitu, Split, Hrvatska

Puni tekst: hrvatski pdf 3.299 Kb

str. 91-98

preuzimanja: 1.050

citiraj

Sažetak

In this paper, k-means algorithm is presented. It is a heuristic algorithm for solving NP-hard optimisation problem of classifying a given data into clusters, with a number of clusters fixed apriori. The algorithm is simple and it's convergence is fast, what makes it widely used, despite its tendency of stoping in a local minimum and inability of recognizing clusters not separated by hyper-planes. The method of the first variation as a tool for escaping from a local minimum is also presented in the paper.

Ključne riječi

k-means algorithm; clustering; fi rst variation method

Hrčak ID:

174106

URI

https://hrcak.srce.hr/174106

Datum izdavanja:

16.1.2017.

Podaci na drugim jezicima: hrvatski

Posjeta: 1.830 *