Please use this identifier to cite or link to this item: http://www.repositorio.ufop.br/handle/123456789/4293
Title: Aggregation Trees for visualization and dimension reduction in many-objective optimization.
Authors: Freitas, Alan Robert Resende de
Fleming, Peter J.
Guimarães, Frederico Gadelha
Keywords: Aggregation trees
Evolutionary computation
Issue Date: 2015
Citation: FREITAS, A. R. R. de.; FLEMING, P. J.; GUIMARÃES, F. G. Aggregation Trees for visualization and dimension reduction in many-objective optimization. Information Sciences, v. 298, p. 288-314, 2015. Disponível em: <http://www.sciencedirect.com/science/article/pii/S0020025514011347#>. Acesso em: 21 jan. 2015.
Abstract: This paper introduces the concept of Aggregation Trees for the visualization of the results of high-dimensional multi-objective optimization problems, or many-objective problems and as a means of performing dimension reduction. The high dimensionality of manyobjective optimization makes it difficult to represent the relationship between objectives and solutions in such problems and most approaches in the literature are based on the representation of solutions in lower dimensions. The method of Aggregation Trees proposed here is based on an iterative aggregation of objectives that are represented in a tree. The location of conflict is also calculated and represented on the tree. Thus, the tree can represent which objectives and groups of objectives are the most harmonic, what sort of conflict is present between groups of objectives, and which aggregations would be helpful in order to reduce the problem dimension.
URI: http://www.repositorio.ufop.br/handle/123456789/4293
ISSN: 00200255
metadata.dc.rights.license: O Periódico Information Sciences concede permissão para depósito do artigo no Repositório Institucional da UFOP. Número da licença: 3552540146011.
Appears in Collections:DECOM - Artigos publicados em periódicos

Files in This Item:
File Description SizeFormat 
ARTIGO_AggregationTreesVsualization.pdf1,94 MBAdobe PDFView/Open


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