Please use this identifier to cite or link to this item:
Type: Relatório
Title: A validity measure for hard and fuzzy clustering derived from Fisher's linear discriminant
Author(s)/Inventor(s): Franco, Cláudia Rita de
Vidal, Leonardo Silva
Cruz, Adriano Joaquim de Oliveira
Abstract: Cluster analysis has a growing importance in many research areas, especially those involving problems of pattern recognition. Generally, in real world problems, the number of classes is unknown in advance, being necessary to have criterions to Identify the best choice of clusters. Here we propose an extension to Fisher Linear Discriminant, the EFLD that does not impose limits on the minimum number of samples, can be applied to fuzzy and crisp partitions and can be calculated more efficiently. We also propose a nem fast and efficient validity method based in the EFLD that measures the compactness and separation of partitions produced by any fuzzy or crisp clustering algorithm. The simulations performed indicate that it's a efficient and fast measure even when the overlapping between clusters is high. Finally, we propose an algorithm that applies the new validity measure to the problem of finding the patterns for the fuzzy K-NN classifier. This algorithm is applied to the problem of cursive digits recognition.
Keywords: Cluster validity
Fuzzy clustering
Pattern recognition
Cursive digits recognition
Fisher's linear discriminant
Agrupamento difuso
Sistemas de reconhecimento de padrões
Production unit: Instituto Tércio Pacitti de Aplicações e Pesquisas Computacionais
In: Relatório Técnico NCE
Issue: 0202
Issue Date: 30-Dec-2002
Publisher country: Brasil
Language: eng
Right access: Acesso Aberto
Citation: FRANCO, C. R. de.; VIDAL, L. S.; CRUZ, A. J. DE O. A validity measure for hard and fuzzy clustering derived from Fischer's linear discriminant. Rio de Janeiro: NCE/UFRJ, 2002. 6 p. (Relatório Técnico, 02/02)
Appears in Collections:Relatórios

Files in This Item:
File Description SizeFormat 
02_02_000613368.pdf532.89 kBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.