M. Bouten et al., GRADIENT DESCENT LEARNING IN PERCEPTRONS - A REVIEW OF ITS POSSIBILITIES, Physical review. E, Statistical physics, plasmas, fluids, and related interdisciplinary topics, 52(2), 1995, pp. 1958-1967
We present a streamlined formalism which reduces the calculation of th
e generalization error for a perceptron, trained on random examples ge
nerated by a teacher perceptron, to a matter of simple algebra. The me
thod is valid whenever the student perceptron can be identified as the
unique minimum of a specific cost function. The asymptotic generaliza
tion error is calculated explicitly for a broad class of cost function
s, and a specific cost function is singled out that leads to a general
ization error extremely close to the one of the Bayes classifier.