Publication details
Clustering by Mode Estimation
Conference Paper (international conference)
serial: IEEE - 7th Slovakian-Hungarian Joint Symposium on Applied Machine Intelligence and Informatics
action: SAMI 2009, (Herlany, SK, 30.01.2009-02.02.2009)
research: CEZ:AV0Z10750506
keywords: clustering, boundary detection, mode boundary estimation
preview: Download
abstract (eng):
In this contribution we introduce a clustering scheme based on mode boundary detection procedures. Modes are characterized as compact regions of the data space with higher densities than their surrounding. A mode boundary as defined in this approach is an area of large local changes in the probability density functions. Examples of the performance of the clustering based on the so-obtained mode boundaries are given using artificially generated data sets.
abstract (cze):
V tomto článku představujeme shlukovací schéma založené na odhadu módu procedur. Módy jsou charakterizovány jako kompaktní části datového prostoru s vyšší hutností než jejich okolí.
RIV: IN