Properties of a binary similarity measure
CDAM Research Reports Series, London, UK.
Here we investigate properties of a measure of similarity between a binary vector and a set of binary vectors that we believe may be useful for classification of medical data. We present combinatorial and asymptotic properties, and some results useful for binary classification. We show that if our underlying function is assumed to be a bounded term DNF, then our hypothesis function will correctly classify any example with large similarity measure.