Title :
Extension of cellular automata to neural computation: theory and applications
Author :
Blayo, François ; Marchal, Pierre
Author_Institution :
Microcomput. Lab., Swiss Fed. Inst. of Lausanne, Switzerland
Abstract :
A relationship between cellular automata and neuro-mimetic theory is presented. A systolic approach is proposed for evaluating a first-neighborhood cellular automaton. This method has been extended for a full-neighborhood automaton and applied to Hopfield-like neural networks. Three possible resulting solutions are discussed, and a new approach which allows for full extensibility is shown. Appropriate external storage of the parameters provides a flexible recognition system able to dynamically modify the synaptic weights. Such a function will be used later for the implementation of an autoadaptive system.<>
Keywords :
finite automata; neural nets; parallel architectures; Hopfield-like neural networks; autoadaptive system; cellular automata; dual ring array architecture; external parameters storage; first neighbourhood automaton; flexible recognition system; full-neighborhood automaton; neural computation; neuro-mimetic theory; synaptic weights; systolic architectures; Finite automata; Neural networks; Parallel architectures;
Conference_Titel :
Neural Networks, 1989. IJCNN., International Joint Conference on
Conference_Location :
Washington, DC, USA
DOI :
10.1109/IJCNN.1989.118699
