Optimality of adaptive algorithms for pattern classifier design

This paper derives an expression for the misadjustment, or extra probability of error, which results when adaptive design algorithms are used to design linear pattern classifiers from a finite number of patterns. The expressions have been evaluated for three algorithms--the fixed-increment error-correction algorithm, the relaxation error-correction algorithm, and the least-mean-square (LMS} algorithm--when used with patterns from Gaussian distributions with equal covariance matrices.