Ilnar Nasyrov
Mikhail Gudyma
Lev Kazakovtsev
Dmitry Stashkov
Reshetnev Siberian State University of Science and Technology
Siberian Federal University
JSC "SINETIC"
Russian Federation
e-mail: levk@bk.ru
Abstract:
In this paper, we propose new algorithms for solving the classical problem of cluster analysis, k-Means, which uses the variable neighbourhoods search in randomized neighbourhoods formed by running the greedy agglomerative procedures. A comparison with known algorithms including the algorithms with greedy heuristics is given, and the advantage is confirmed experimentally.
Key words:
clustering genetic algorithm
greedy agglomerative heuristic
k-means
p-median
Section:
Topics: