Minimum Distance
Overview
The minimum distance algorithm is used in classification problems where one has a set of
distributions representing the various categories in the problem and also a data point where one
wishes to determine which distribution that data point was drawn from.
In such a case, if one has a distance function (similar to a
Metric),
which measures the distance between a point and a distribution, then one simply finds the distribution that is
"closest" to the data point, as specified by the distance function.
Note: there is a distinction between the concept of a metric in a topological sense, and the type of metric
used here. In this case, the distance function measures the distance between two unlike objects,
a data point and a distribution. It can be re-interpreted in terms of a metric or pseudo-metric, particularly
if the distribution is represented by some known population.
Measures
