Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Automatic kernel clustering with bee colony optimization algorithm
Kuo R., Huang Y., Lin C., Wu Y., Zulvia F. Information Sciences283 107-122,2014.Type:Article
Date Reviewed: Apr 8 2015

Clustering is an unsupervised classification approach extensively used in data mining and pattern recognition. Many clustering techniques have been developed so far. The value of each algorithm depends on the particular problem where it is applied, and some problems might require the automatic estimation of the number of clusters. The algorithms for establishing the number of clusters, in up-to-date designs, are usually based on the measure of the within-cluster variance.

This paper proposes a clustering method that automatically predicts an optimum number of target clusters. The novelty and strength of the algorithm is the fact that it operates with two state-of-the-art concepts, the bee colony optimization and kernel functions, known to have significant discriminative power. Another advantage of the proposed method is the fact that the number of clusters is determined during the clustering process. The clustering performance is estimated as the tradeoff among the within-cluster and between-cluster variances, calculated based on a specific Gaussian kernel function. It is used as an objective function for the bee-colony optimization algorithm. Certain clusters are discarded based on a threshold value to finally generate the suitable number of clusters. The algorithm, called automatic kernel clustering with bee colony optimization (AKC-BCO), is presented in the paper in a somewhat metaphorical language; it would have been more helpful if the authors had provided details for some steps, such as the parameter setup in step 1, or step 3 (employed bees randomly generate the location of each food source). Moreover, the flowchart, presented in figure 1, expresses the same information as the algorithm using the same language.

Several experimental results are presented in the paper, provided either on benchmark datasets or on serious real-life problems such as prostate cancer diagnosis. The algorithm is compared to some other swarm optimization methods. The positive conclusions are sustained by statistical t-tests that validate the benefits of the AKC-BCO algorithm. Yet, in all of the examined cases, the number of clusters is somehow known in advance. A really unsupervised case study, where the number of clusters is discovered, would be interesting. However, the structure of the experiments is well thought out and confirms the value of the paper.

Reviewer:  Svetlana Segarceanu Review #: CR143322 (1507-0613)
Bookmark and Share
 
Clustering (H.3.3 ... )
 
 
Data Mining (H.2.8 ... )
 
 
Pattern Recognition (I.5 )
 
Would you recommend this review?
yes
no
Other reviews under "Clustering": Date
Concepts and effectiveness of the cover-coefficient-based clustering methodology for text databases
Can F. (ed), Ozkarahan E. ACM Transactions on Database Systems 15(3): 483-517, 1990. Type: Article
Dec 1 1992
A parallel algorithm for record clustering
Omiecinski E., Scheuermann P. ACM Transactions on Database Systems 15(3): 599-624, 1990. Type: Article
Nov 1 1992
Organization of clustered files for consecutive retrieval
Deogun J., Raghavan V., Tsou T. ACM Transactions on Database Systems 9(4): 646-671, 1984. Type: Article
Jun 1 1985
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy