
7.6 Miscellaneous Clustering Algorithms 189
7.6 MISCELLANEOUS CLUSTERING ALGORITHMS
In this section, we consider algorithms that produce a single clustering and do not fall into either the
sequential or the cost function optimization category.
Competitive Leaky Learning Algor ithm
Competitive leaky learning (LLA) is an algorithm suitable for unraveling compact clusters. Once again,
the number of clusters, m, that underlie the data set X, is assumed to be known. The aim of LLA is to
move ml-dimensional parameter vectors, w
j
’s, j = 1, ..., m,
5
to regions that are “dense” in points of
X. Each parameter vector represents one “dense” region (cluster). Th ...