Naslov
Classification of bivariate distribution based on EM algorithm

Autori
Barbir, Ante ; Ban, Dario ; Perišić, Stipe

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Proceedings of the 9th International Congress of Croatian Society of Mechanics / - Zagreb : Hrvatsko društvo za mehaniku (HDM), 2018

Skup
9th International Congress of Croatian Society of Mechanics (ICCSM 2018)

Mjesto i datum
Split, Hrvatska, 18.09.2018. - 22.09.2018

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
classification ; estimation-maximization EM algorithm ; radial basis functions

Sažetak
In present, there are many possibilities to collect large amount of data readings during measurement of specific physical quantities on various systems. This large number of readings is time- consuming for determining in which class some data could be assigned. To reduce this classification time, system which classifies reading parameters using estimation- maximization algorithm can be introduced, and afterwards using radial basis functions, get measuring results. EM algorithm is used on two generated bivariate distributions to make classification, and RBF interpolation has been made to estimate third dimension of data. Third dimension can represent various measurement results including incomplete data readings, which can be restored using RBF interpolation due RBF reconstruction property. Two sets of Gaussian distribution were used in this paper, classified, and interpolated using radial basis functions for the multivariate interpolation with Euclidean L2 norm.

Izvorni jezik
Engleski

Znanstvena područja
Strojarstvo

POVEZANOST RADA