Pattern analysis of Békésy audiograms.

A data-dependent method for the objective classification of Békésy audiograms is introduced. The technique uses orthogonal transformation from the pattern space to a reduced dimensionality feature space that allows a classification process to be implemented digitally. It is argued that this technique, with accretion of audiograms, could form the basis of a data-based system for determining and classifying the coherent structures of the various Békésy patterns with as few as possible a priori hypotheses made about the Békésy "types."