Регистрация / Вход
Прислать материал

Adaptive Weights Clustering

Name
Spokoiny
Surname
Vladimir
Scientific organization
Skoltech, MIPT, WIAS and HU Berlin
Academic degree
Prof.
Position
Prof., Head of research group
Scientific discipline
Information technologies
Topic
Adaptive Weights Clustering
Abstract
This paper presents a new non-parametric method of cluster analysis called Adaptive Weights Clustering.
The proposed procedure aims at describing the clustering structure in terms of weights w_{ij} which is computed from
the data by an iterative procedure.
The method is fully adaptive and does not require to specify the number of clusters or their structure.
The procedure is able to recover different clusters with sharp edges or manifold structure. A numerical study illustrates a nice performance of the method.
Keywords
Cluster gap, propagation, separation
Summary

"No"