SOLUTION OF AUTOMATIC SELECTION OF ALGORITHM FRIS-CLUSTER PARAMETER R*
Zyryanov Aleksandr Olegovich, Pavlovskii Evgenii Nikolaevich, Dyubanov Vladimir Vladimirovich
Novosibirsk State University
Novosibirsk
Abstract. The work is devoted to the clustering algorithm FRiS-Cluster, the weak point of which is the parameter r*. This parameter must be specified manually, thus the correctness of choice can be judged only at the end of the algorithm. The article explores the dependence of the quality of clustering task solution on the value of r*. The heuristics for determining the obviously poor quality of clustering before the end of the algorithm working is formulated. On the basis of the derived heuristics an algorithm for automatic selecting the parameter r* is proposed.
Key words and phrases: кластеризация, кластерный анализ, таксономия, FRiS-методология, столп, когнитивный анализ данных, виртуальный противник, расстояние до виртуального противника, clustering, cluster analysis, taxonomy, FRiS-methodology, pillar, cognitive analysis of data, virtual opponent, distance to virtual opponent
Open the whole article in PDF format. Free PDF-files viewer can be downloaded .
References:
Zagoruiko N. G. Kognitivnyi analiz dannykh. Novosibirsk: Geo, 2013.
Borisova I. A., Dyubanov V. V., Kutnenko O. A., Zagoruiko N. G. Use of the FRiS-Function for Taxonomy, Attribute Selection and Decision Rule Construction // Lecture Notes in Artificial Intelligence. 2011. V. 6581. P. 256-270.
Zagoruiko N. G., Borisova I. A., Dyubanov V. V., Kutnenko O. A. A Quantitative Measure of Compactness and Similarity in a Competitive Space // Journal of Applied and Industrial Mathematics. 2011. Vol. 5. № 1. P. 144-154. DOI: 10.1134/S1990478911010157.
Zagoruiko N. G., Borisova I. A., Dyubanov V. V., Kutnenko O. A. Methods of Recognition Based on the Function of Rival Similarity // Pattern Recognition and Image Analysis. 2008. Vol. 18. № 1. P. 1-6. DOI: 10.1007/s11493-008-1001-8.
Русский
