This paper describes a genetically guided approach to optimizing the hard (
J(1)) and fuzzy (J(m)) c-means functionals used in cluster analysis. Our ex
periments show that a genetic algorithm (GA) can ameliorate the difficulty
of choosing an initialization for the c-means clustering algorithms. Experi
ments use six data sets, including the Iris data, magnetic resonance, and c
olor images. The genetic algorithm approach is generally able to find the l
owest known. J(m) value or a J(m) associated with a partition very similar
to that associated with the lowest. J(m) value. On data sets with several l
ocal extrema, the GA approach always avoids the less desirable solutions. D
egenerate partitions are always avoided by the GA approach, which provides
an effective method for optimizing clustering models whose objective functi
on can be represented in terms of cluster centers. A series random initiali
zations of fuzzy/hard c-means, where the partition associated with the lowe
st J(m) value is chosen, can produce an equivalent solution to the genetic
guided clustering approach given the same amount of processor time in some
domains.