FAULT-TOLERANT DESIGN OF NEURAL NETWORKS FOR SOLVING OPTIMIZATION PROBLEMS

First, the incorporation of the alternative ways of representing the s olution to the traveling salesman problem by neural network is propose d to deal with the asymmetric tolerance of neural networks to unidirec tional faults. Next, in order to cope with the simultaneous occurrence of a s-a-O and s-a-1 faults, the triplication scheme with a proper me rging function is considered. Finally, the combination of the alternat ive representations and the triplication scheme is used.