Carleton University
Technical Report TR-00-03
May 2000
On Optimal Pairwaise Linear Classifiers for Normal Distributions: The Two-Dimensional case
Luis Rueda & B. John Oommen
Abstract
Computing linear classi fiers is a very important problem in statistical Pattern Recognition (PR). These classifi ers have been investigated by the PR community extensively since they are the ones which are both easy to implement and comprehend. It is well known that when dealing with normally distributed classes, the optimal discriminant function for two-classes is linear only when the covariance matrices are equal. Other approaches, such as the Fisher’s discriminant, the perceptron algorithm, minimum square distance classi fiers, etc., have solved this problem by generating a linear classifi er in normal and non-normal distributions, but these classi fiers are typically suboptimal.