A Credal Approach to Naive Classification
Marco Zaffalon · International Symposium on Imprecise Probabilities and Their Applications · 1999
Convex sets of probability distributions are also called credal sets. They generalize probability theory with special regard to the relaxation of the precision requirement about the probability values. Classification, i.e., assigning class labels to instances described by a set of attributes, is a typical domain of application of Bayesian methods, where the naive Bayesian classifier is considered among the best tools. This paper explores the classification model, called naive credal classifier, obtained when the naive Bayesian classifier is extended to credal sets. A fast classification algorithm is derived. A data-driven construction of the classifier is proposed and discussed. The latter takes the variability of the probability estimates into account, thus making classification more reliable.