One of the simplest classifier is the linear discriminant function as follows:
This function is a linear function of the components of X. A complete specification of this linear discriminant function is achieved by specifying the values of the weights or parameters of the function family. The training process is one of adjusting the coefficients ( , , , , ) so that the decision surface implements an acceptable separation of the two classes of patterns in the training set. The decision surfaces of any pattern classifier can be implicitly defined by a set of linear discriminant functions containing R members. Let , , , be scalar and single-valued functions of the pattern X. These functions are called discriminant functions.