Connectionists: JNCC2: an open source classifier based on imprecise probabilities

[Giorgio Corani]

JNCC2 is the Java implementation of the Naive Credal Classifier 2
(Corani and Zaffalon, 2007). NCC2
constitutes an extension of Naive Bayes  towards imprecise
probabilities; it is designed to return robust classification, even on
small and/or incomplete data sets. A peculiar feature of NCC2 is that it
returns set-valued (or imprecise) classifications (i.e., more than one
class) when faced with doubtful instances. Imprecise classifications are
valuable as they clearly highlight doubtful instances, preventing
over-confident use of the issued judgments; however, they still convey
an informative content, dropping unlikely classes. This could be
appealing in contexts in which the classification outcome is especially
sensitive, as for instance in medical area.

Extensive empirical investigation shows that NCC2 returns imprecise
judgments on instances whose classification is truly  doubtful; in fact,
Naive Bayes  achieves a much higher classification accuracy on the
instances precisely classified by NCC2, than on those imprecisely
classified by NCC2. We say that NCC2 isolates area of ignorance, i.e.
subsets of instances over which the accuracy of Naive Bayes sharply drops.

JNCC2 is open source; it is released under the terms of the GNU GPL
license; it is hence freely available together with manual, sources and
javadoc documentation.

For more information, visit http://www.idsia.ch/~giorgio/jncc2.html.