Abstract:
Using the combinatorial-logical approach, we consider the problem
of clustering a set of vectors from an $n$-dimensional vector space of
attributes.
The main results of the paper are the study of the optimal partitions
into clusters with respect to some functionals and the design of iterative
and gradient algorithms of clustering based on this consideration.
It is shown that all optimal partitions are well-separable with respect to the
corresponding decision rules.
Fulltext:
PDF file (1502 kB)