A new approach to weighted fuzzy production rule extraction from neural networks

There are many advantages of artificial neural networks such as high prediction accuracy, robustness, no requirements on data distribution, but knowledge captured by neural networks is not transparent to users. This results in a major problem for users of neural network-based systems. It is significant to extract rules from neural networks. This paper proposes a new method for extracting weighted fuzzy production rules from trained neural networks by structural learning based on matrix of importance index.