Detail publikace
Exploring k-PSO Algorithm for Clustering
HERMAN, D. ORSÁG, F.
Originální název
Exploring k-PSO Algorithm for Clustering
Anglický název
Exploring k-PSO Algorithm for Clustering
Jazyk
en
Originální abstrakt
Cluster analysis is a very popular approach to fully automatic search for patterns, data concepts, groups and clusters. It simplifies data representations and thus plays an important role in the process of knowledge acquisition. Data mining tasks require fast and accurate partition of data with many attributes. This requires new approach, which could deal better with these features. Methods based on the swarm intelligence present such approach to the cluster analysis. This article is a brief introduction to the optimization algorithms inspired by the natural world. It shows how these algorithms can be used in the cluster analysis and describes several up-to-date hybrid techniques combining PSO and k-means. Moreover, conceptually new hybrid algorithm based on the PSO and k-means is introduced and its efficiency and robustness are compared to the other algorithms using several datasets.
Anglický abstrakt
Cluster analysis is a very popular approach to fully automatic search for patterns, data concepts, groups and clusters. It simplifies data representations and thus plays an important role in the process of knowledge acquisition. Data mining tasks require fast and accurate partition of data with many attributes. This requires new approach, which could deal better with these features. Methods based on the swarm intelligence present such approach to the cluster analysis. This article is a brief introduction to the optimization algorithms inspired by the natural world. It shows how these algorithms can be used in the cluster analysis and describes several up-to-date hybrid techniques combining PSO and k-means. Moreover, conceptually new hybrid algorithm based on the PSO and k-means is introduced and its efficiency and robustness are compared to the other algorithms using several datasets.
Dokumenty
BibTex
@inproceedings{BUT103428,
author="David {Herman} and Filip {Orság}",
title="Exploring k-PSO Algorithm for Clustering",
annote="Cluster analysis is a very popular approach to fully automatic search for
patterns, data concepts, groups and clusters. It simplifies data representations
and thus plays an important role in the process of knowledge acquisition. Data
mining tasks require fast and accurate partition of data with many attributes.
This requires new approach, which could deal better with these features. Methods
based on the swarm intelligence present such approach to the cluster analysis.
This article is a brief introduction to the optimization algorithms inspired by
the natural world. It shows how these algorithms can be used in the cluster
analysis and describes several up-to-date hybrid techniques combining PSO and
k-means. Moreover, conceptually new hybrid algorithm based on the PSO and k-means
is introduced and its efficiency and robustness are compared to the other
algorithms using several datasets.",
address="ACTA Press",
booktitle="Proceedings of the IASTED International Conference Artificial Intelligence and Applications (AIA 2013)",
chapter="103428",
edition="NEUVEDEN",
howpublished="electronic, physical medium",
institution="ACTA Press",
year="2013",
month="february",
pages="161--168",
publisher="ACTA Press",
type="conference paper"
}