Towards Nature-Inspired Modularization of Artificial Neural Networks via Static and Dynamic Weights

Abstract

Many conventional artificial neural network (ANN) models are designed for one application domain only. The work presented in this paper describes ANN models that operate with a higher economy by sharing neurons across domains. The use of two different types of weights-static weights and dynamic weights-is a fundamental feature of the presented models. Results from a comprehensive series of experiments provide evidence for the meaningfulness of the proposed models. © Springer-Verlag Berlin Heidelberg 2014.