An execution system of logic programming language using neural networks-an improvement of the transformation algorithm

We propose an execution system for a logic programming language using neural networks. We transform the style like logic programming language into a Hopfield-type neural network and attempt to execute the logic programming language using the ability of the neural network as an optimization machine. Although we have proposed such a system, previous works have not implemented "list processing". Therefore we propose an improved transformation algorithm for "list processing". The performance of the new transformation algorithm is evaluated by comparing it with a previous algorithm, without handling list structure, for a logic programming language. The result shows that the proposed algorithm generates an especially smaller network scale than a conventional algorithm, and reduces iteration times.