Abstract
NEXTOOL combines a hybrid scheme for knowledge representation that seems to be a powerful and flexible tool for developing heuristic classification systems (Machado et al, 1991b). It combines the expressiveness of semantic networks, the naturalness of fuzzy logic and the learning power of both inductive and deductive learning strategies.
The semantic networks give the system the ability to represent symbolic concepts, to structure and organize the problem domain knowledge, and to provide high level inference mechanisms such as the choice of the best reasoning models to solve a particular task.
The learning capability provided by the inductive and the deductive strategies supply a very potent tool to make of artificial intelligent systems structures very adaptive to a changing environment. Some interesting capabilities of such systems are: learning from scratch; automatic conversion of external (expert graphs) knowledge into EKN; continuous knowledge refinement, etc.
The inference and inquiry processes are low cost processes in NEXTOOL because they are supported by local decision and acyclic networks. Also, the description of NEXTOOL decisions is a very natural set of fuzzy rules of the type if X is A and Y is B .... then Z is C because its reasoning is supported by fuzzy logics.
Preview
Unable to display preview. Download preview PDF.
Rights and permissions
Copyright information
© 1992 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
(1992). NEXTOOL: A MPNN classifying system. In: Neural Nets. Lecture Notes in Computer Science, vol 638. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-55949-3_17
Download citation
DOI: https://doi.org/10.1007/3-540-55949-3_17
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-55949-8
Online ISBN: 978-3-540-47317-6
eBook Packages: Springer Book Archive