This article points out some very serious misconceptions about the brain in connectionism and artificial neural networks. Some of the connectionist ideas have been shown to have l...
This paper describes the methodology of developing multiversion systems using neural networks in the hope of improving their performance and reliability. However, a system impleme...
In this paper, a recurrent neural network is used to develop a dynamic controller for mobile robots. The advantage of the control approach is that no knowledge about the robot mode...
Mohamed Oubbati, Michael Schanz, Thorsten Buchheim...
In the `missing data' approach to improving the robustness of automatic speech recognition to added noise, an initial process identifies spectraltemporal regions which are do...
This paper presents an algorithm for extract ing propositions from trained neural networks. The algorithm is a decompositional approach which can be applied to any neural networ...