Efficient recurrent neural network training incorporating a priori knowledge

A new technique for efficient training of Hopfield network models using iterative training algorithms is described and demonstrated. This technique is useful for producing stable Hopfield networks, using recently derived results concerning stability conditions for the Hopfield network. Since the modified training algorithm ensures the stability of the network, no off-line verification of stability is required. This technique can be applied to Hopfield networks of any size, and therefore is tested for three different randomly selected sizes. Additionally, the advantage of utilising a priori information about the plant is also tested and the results are compared with those cases where no such information is available. This information is used to determine critical dynamic properties of the network model, necessary for the network’s ability to generalise. In all tests, the networks are trained with modified genetic algorithms, using different initial starting points for the algorithm.