Iterated learning in a layered feed-forward neural network
Ronny Meir, Eytan Domany · Physical Review A · 1988
A model-layered feed-forward neural network with an iterated learning stage is described. The system is capable of self-organizing its ``internal representations'' so that a large number of patterns can be perfectly stored and retrieved. The performance of the network is found to be robust against various degrading operations. Comparison of performance with an exactly solvable model is made.