RT info:eu-repo/semantics/article
T1 Improving the computational performance of TCLUST through ensemble initialization
A1 Álvarez Esteban, Pedro César
A1 García Escudero, Luis Ángel
A1 Mayo Iscar, Agustín
A1 Crespo Guerrero, Javier
K1 Cluster analysis
K1 Robustness
K1 Trimming
K1 Model-based clustering
K1 12 Matemáticas
AB Outliers are known to be detrimental to widely used clustering techniques. Robust clus-tering alternatives have been introduced to better resist outlying observations. Amongthese, robust clustering methods based on trimming have proven effective by allowingthe removal of a fraction of observations where outliers are likely to be found, withTCLUST being one of the most popular for handling elliptically contoured clusters.The algorithm for applying TCLUST can be seen as an extension of the concentra-tion steps used in the fast-MCD algorithm for computing the Minimum CovarianceDeterminant. However, obtaining good initializations for these concentration steps inTCLUST is more complex than in MCD. This initialization task is particularly chal-lenging unless both the number of clusters and the dimensionality are small. To addressthis, a new ensemble initialization procedure for TCLUST will be presented, whichtakes advantage of partially correct information from all iterated random initializa-tions rather than focusing solely on the best individual one found. Initial experimentssuggest that this methodology could improve the computational performance of thestandard TCLUST algorithm.
PB Springer
SN 1862-5347
YR 2025
FD 2025
LK https://uvadoc.uva.es/handle/10324/75941
UL https://uvadoc.uva.es/handle/10324/75941
LA eng
NO Advances in Data Analysis and Classification, 2025.
NO Producción Científica
DS UVaDOC
RD 14-jun-2025