TY - JOUR
T1 - Knowledge extraction from meural networks using the all-Permutations fuzzy rule base
T2 - The LED display recognition problem
AU - Kolman, Eyal
AU - Margaliot, Michael
PY - 2007/5
Y1 - 2007/5
N2 - A major drawback of artificial neural networks (ANNs) is their black-box character. Even when the trained network performs adequately, it is very difficult to understand its operation. In this letter, we use the mathematical equivalence between ANNs and a specific fuzzy rule base to extract the knowledge embedded in the network. We demonstrate this using a benchmark problem: the recognition of digits produced by a light emitting diode (LED) device. The method provides a symbolic and comprehensible description of the knowledge learned by the network during its training.
AB - A major drawback of artificial neural networks (ANNs) is their black-box character. Even when the trained network performs adequately, it is very difficult to understand its operation. In this letter, we use the mathematical equivalence between ANNs and a specific fuzzy rule base to extract the knowledge embedded in the network. We demonstrate this using a benchmark problem: the recognition of digits produced by a light emitting diode (LED) device. The method provides a symbolic and comprehensible description of the knowledge learned by the network during its training.
KW - Feedforward neural networks
KW - Hybrid intelligent systems
KW - Knowledge extraction
KW - Neurofuzzy systems
KW - Rule extraction
KW - Rule generation
UR - http://www.scopus.com/inward/record.url?scp=34248681042&partnerID=8YFLogxK
U2 - 10.1109/TNN.2007.891686
DO - 10.1109/TNN.2007.891686
M3 - ???researchoutput.researchoutputtypes.contributiontojournal.article???
AN - SCOPUS:34248681042
SN - 2162-237X
VL - 18
SP - 925
EP - 931
JO - IEEE Transactions on Neural Networks and Learning Systems
JF - IEEE Transactions on Neural Networks and Learning Systems
IS - 3
ER -