Home // International Journal On Advances in Intelligent Systems, volume 8, numbers 1 and 2, 2015 // View article

Decision Support System for Neural Network R&D

Authors:
Rok Tavčar
Jože Dedič
Andrej Žemva
Drago Bokal

Keywords: Neural Networks; DSS; Knowledge Base; Taxonomy; Inference Engine.

Abstract:
One of the reasons that keep Neural Networks (NNs), which are advanced computational methods (ACM) of great potential, from coming into broader practical use, is the lack of systematic method in finding the optimal match between NN architecture and target application. If this match is performed erratically, practical solutions often yield unimpressive results. It is the a) validation of the problem's fitness for a NN-based solution and b) matching of an optimal NN implementation to the given problem that is crucial. This paper presents a theoretical foundation for an inference engine decision space and a taxonomic framework for a knowledge base, which are part of our proposed knowledge-driven decision support system (DSS). Furthermore, this paper provides details of our inference engine, namely a) an algorithm for optimal matching of a NN setup against the given learning task, b) the application of this same algorithm for interactive exploration of the decision space and c) an algorithm for automatic inference of potential NN research synergies based on existing successful NN solutions. Finally, we propose a process for establishing and maintaining the growth of the DSS knowledge database.

Pages: 182 to 193

Copyright: Copyright (c) to authors, 2015. Used with permission.

Publication date: June 30, 2015

Published in: journal

ISSN: 1942-2679