The Complete SPRLIB & ANNLIB

pr_error_multi

- compute the classification error for a classifier

SYNOPSIS

double pr_error_multi (K, M, W, b, n)

ARGUMENTS

double **K The covariance matrix.
double *M The mean vector of size [1..n].
double *W The weight vector of size [1..n].
double *b The constant vector.
int n The dimensionality.

RETURNS

If an error was detetected the function returns a -1.0, otherwise it returns the classification error made.

FUNCTION

The function computes the classification error of a n-dimensional classifier given by F(x) = W*x + b. The routine calculates the probability of a random pattern for which F(x) < 0 assuming a normal distribution.

NOTE

The variables b, W and M are in vector format and K is in matrix format.

SEE ALSO

pr_error

This document was generated using api2html on Thu Mar 5 09:00:00 MET DST 1998