A classifier which, in general, implements a nonlinear decision boundary is shown to be equivalent to a linear discriminant function when the measurements are binary valued; its relation to the Bayes classifier is derived. The classifier requires less computation than a similar one based on the Euclidean distance and can perform equally well.
|Original language||English (US)|
|Number of pages||3|
|Journal||IEEE Transactions on Systems Science and Cybernetics|
|State||Published - Oct 1970|
ASJC Scopus subject areas