Iterative merging heuristics for correlation clustering
Författare
Summary, in English
A straightforward natural iterative heuristic for correlation clustering in the general setting is to start from singleton clusters and whenever merging two clusters improves the current quality score merge them into a single cluster. We analyse the approximation and complexity aspects of this heuristic and its three simple deterministic or random refinements.
Avdelning/ar
- Computer Science
- Matematik (naturvetenskapliga fakulteten)
Publiceringsår
2014
Språk
Engelska
Sidor
105-117
Publikation/Tidskrift/Serie
International Journal of Metaheuristics
Volym
3
Issue
2
Dokumenttyp
Artikel i tidskrift
Förlag
Inderscience Publishers
Ämne
- Computer Science
Nyckelord
- Randomised algorithms
- Time complexity
- Approximation algorithms
- Correlation clustering
- Graph clustering
Status
Published
ISBN/ISSN/Övrigt
- ISSN: 1755-2184