Distance classifier

from Wikipedia, the free encyclopedia

A distance classifier is a term from mathematical statistics : If a decision is made as to whether an object belongs to a class by determining the "distance" of the object (e.g. a characteristic vector ) to the class (in the feature space ) , one speaks of a distance classifier .

The distance is usually defined as the distance to a selected representative of each class, for example to its center point . The distance can be defined using different distance or similarity measures . Frequently used distance measures are, for example, the Euclidean distance , the weighted Euclidean distance and the Mahalanobis distance .

The dividing surfaces between two classes form the points equidistant from each of the two classes.

See also