Extracting rules from neural networks using symbolic algorithms: preliminary results

Although Artificial Neural Networks (ANNs) have been satisfactorily employed in several problems, such as clustering, pattern recognition, dynamic systems control and prediction, they still suffer from significant limitations. One of them is that the induced concept representation is not usually comprehensible to humans. Several techniques have been suggested to extract meaningful knowledge from trained ANNs. This paper proposes the use of symbolic learning algorithms, commonly used by the Machine Learning community, to extract symbolic representations from trained ANNs. The procedure proposed is similar to that used by the Trepan algorithm (Craven, 1996), which extracts comprehensible, symbolic representations (decision trees) from trained ANNs