Algorithm with Evenly Distributed Gaussian Function for Diatom Classification

Naumoski, Andreja and Mitreski, Kosta (2010) Algorithm with Evenly Distributed Gaussian Function for Diatom Classification. In: Proceedings of the Seventh Conference on Informatics and Information Technology. Institute of Informatics, Faculty of Natural Sciences and Mathematics, Ss. Cyril and Methodius University in Skopje, Macedonia, Skopje, Macedonia, pp. 28-32. ISBN 978-9989-668-88-3

[img]
Preview
Text
978-9989-668-88-3_pp28-32.pdf

Download (303kB) | Preview
Official URL: http://ciit.finki.ukim.mk

Abstract

The diatom organisms are good bio-indicators of certain ecosystem environments. According the national directive for water quality classification, each WQC represent a water quantity of certain physico-chemical parameters in certain range define by biological experts. The property of bioindicator is used to characterize the environment and thus helping in process of classification of the diatoms in the correct water quality classes (WQCs). In this direction we use pattern trees; trees which have combined the advantages of the information theory and fuzzy theory to model (predict) in which WQC belongs the certain diatom. Because many of the newly discover diatoms does not have ecological preference, this algorithm significantly improves the process of fast and accuracy classification. In our approach we divide each diatom into three evenly ranges with Gaussian functions, which will be represented with fuzzy terms (low, medium and high) similar as the WQC range classes. Using this data mining techniques we can closely reflect the very nature of the diatoms dataset, which later the experiments will confirms this assumption, by taking into account the mean and the standard deviation of each diatom range. The experimental results have shown that the extract knowledge has high level of confidence factor in many cases and the trees obtained have high accuracy compared with other classification algorithms. As future work we intend to expand the number of fuzzy membership and inspect their influence, to implement more fuzzy aggregation functions and similarity definitions in process of pattern trees.

Item Type: Book Section
Subjects: International Conference on Informatics and Information Technologies > Artificial Intelligence
International Conference on Informatics and Information Technologies > Robotics
International Conference on Informatics and Information Technologies > Bioinformatics
Depositing User: Vangel Ajanovski
Date Deposited: 28 Oct 2016 00:15
Last Modified: 28 Oct 2016 00:15
URI: http://eprints.finki.ukim.mk/id/eprint/11037

Actions (login required)

View Item View Item