Using genetic algorithms to extract rules from trained neural networks
Edward C. Keedwell, Ajit Narayanan, Dragan A. Savic · 1999
This paper describes a novel method of extracting rules from trained artificial neural networks. The system which uses genetic algorithms to find an optimal path through the network for a classification. It is claimed that this system compares favourably to traditional rule extraction algorithms due to its ability to extract rules from very large and complex networks without an exponential increase in computational complexity. 1