Artificial neural networks; definition, properties and misuses

Abstract

There are no such clear and good definitions of ANNs in the literature. Many of the definitions refer to the figures instead of well explaining the ANNs. That is why many weighted graphs (as in shortest path problem networks) fit the definition of ANN. This study aims to give a clear definition that will differentiate ANN and graphs (or networks) by referring to biological neural networks. Although there is no input choice limitation or prior assumption in ANN, sometimes researchers compare ANN achievements with the results of other methods using different input data and make comments on these results. This study also gives examples from literature to misuses, unfair comparisons and evaluates the underlying reasons which will guide researchers. © 2011 Nova Science Publishers, Inc. All rights reserved.