We study the problem of determining an optimal bipartition {A,B} of a set X of n points in ℝ^2 that minimizes the sum of the sample variances of A and B, under the size constraints |A| = k and |B| = n − k. We present two algorithms for such a problem. The first one computes the solution in O(n\sqrt[3]{k}\log^2 n) time by using known results on convex-hulls and k-sets. The second algorithm, for an input X ⊂ ℝ^2 of size n, solves the problem for all k = 1,2,..., n/2 and works in O(n^2 logn) time.
Exact algorithms for 2-clustering with size constraints in the Euclidean plane / A. Bertoni, M. Goldwurm, J. Lin (LECTURE NOTES IN COMPUTER SCIENCE). - In: SOFSEM 2015 : theory and practice of computer science / [a cura di] G.F. Italiano, T. Margaria-Steffen, J. Pokorny, J.J. Quisquater, R. Wattenhofer. - Prima edizione. - Heidelberg : Springer, 2015. - ISBN 9783662460771. - pp. 128-139 (( Intervento presentato al 41. convegno International Conference on Current Trends in Theory and Practice of Computer Science tenutosi a Pec pod Sněžkou nel 2015 [10.1007/978-3-662-46078-8_11].
Exact algorithms for 2-clustering with size constraints in the Euclidean plane
A. BertoniPrimo
;M. GoldwurmSecondo
;J. LinUltimo
2015
Abstract
We study the problem of determining an optimal bipartition {A,B} of a set X of n points in ℝ^2 that minimizes the sum of the sample variances of A and B, under the size constraints |A| = k and |B| = n − k. We present two algorithms for such a problem. The first one computes the solution in O(n\sqrt[3]{k}\log^2 n) time by using known results on convex-hulls and k-sets. The second algorithm, for an input X ⊂ ℝ^2 of size n, solves the problem for all k = 1,2,..., n/2 and works in O(n^2 logn) time.File | Dimensione | Formato | |
---|---|---|---|
biclustereuclideo_finale.pdf
accesso riservato
Descrizione: Articolo completo
Tipologia:
Post-print, accepted manuscript ecc. (versione accettata dall'editore)
Dimensione
167.85 kB
Formato
Adobe PDF
|
167.85 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.