The BRAINNE method is a technique for extracting symbolic production rules, concepts and concept hierarchies from neural networks. Previous work reported on the extraction of conjunctive and disjunctive rules for the case of binary, discrete and continuous features. In this paper we explain three new improvements. The first improvement uses a hybrid two layer network for learning disjunctive rules, where the first layer consists of a network that uses unsupervised learning and the second layer uses supervised learning. The second improvement examines the effect on the generalisation capability of the rules developed by avoiding overtraining of the neural network, and the third development is an improved approach to dealing with continuous features which greatly increases the generalisation capability of the derived rules.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.