The self-organizing map (SOM) method is a new, powerful software teal
for the visualization of high-dimensional data. It concerts complex, n
onlinear statistical relationships between high-dimensional data into
simple geometric relationships on a low-dimensional display. As if the
reby compresses information while preserving the most important topolo
gical and metric relationships of the primary data elements on the dis
play, it may also be thought ro produce some kind of abstractions. The
se two aspects, visualization and abstraction, occur in a number of co
mplex engineering tasks such as process analysis, machine perception,
control, and communication. The term self-organizing map signifies a c
lass of mappings defined by error-theoretic consideration. In practice
they result in certain unsupervised, competitive learning processes,
computed by simple-looking SOM algorithms, The first SOM algorithms we
re conceived around 1981-1982, and the popularity of the more advanced
SOM methods is growing at a steady pace. Many industries have found t
he SOM-based software tools useful. The most important property of the
SOM, orderliness of the input-output mapping, can be utilized for man
y tasks: reduction of the amount of training data, speeding up learnin
g, nonlinear interpolation and extrapolation, generalization, and effe
ctive compression of information for its transmission.