Abstract
This work describes a methodology to extract symbolic rules from trained neural networks. In our approach, patterns on the network are codified using formulas on a Łukasiewicz logic. For this we take advantage of the fact that every connective in this multi-valued logic can be evaluated by a neuron in an artificial network having, by activation function the identity truncated to zero and one. This fact simplifies symbolic rule extraction and allows the easy injection of formulas into a network architecture. We trained this type of neural network using a back-propagation algorithm based on Levenderg-Marquardt algorithm, where in each learning iteration, we restricted the knowledge dissemination in the network structure. This makes the descriptive power of produced neural networks similar to the descriptive power of Łukasiewicz logic language, minimizing the information loss on the translation between connectionist and symbolic structures. To avoid redundance on the generated network, the method simplifies them in a pruning phase, using the “Optimal Brain Surgeon” algorithm. We tested this method on the task of finding the formula used on the generation of a given truth table. For real data tests, we selected the Mushrooms data set, available on the UCI Machine Learning Repository.
We gratefully acknowledge the support of the Instituto Superior de Engenharia de Lisboa and the Área Cientifica da Matemática. This paper benefited from the valuable comments of ICFC2009 conference referees.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Hitzler, P., Hölldobler, S., Seda, A.K.: Logic programs and connectionist networks. Journal of Applied Logic 2, 245–272 (2004)
d’Avila Garcez, A.S.: Advances in neural-symbolic learning systems: Modal and temporal reasoning. In: Hammer, B., Hitzler, P. (eds.) Perspectives of Neural-Symbolic Integration. SCI, vol. 77. Springer, Heidelberg (2007)
Komendantskaya, E., Lane, M., Seda, A.K.: Connectionistic representation of multi-valued logic programs. In: Hammer, B., Hitzler, P. (eds.) Perspectives of Neural-Symbolic Integration. SCI, vol. 77. Springer, Heidelberg (2007)
Eklund, P., Klawonn, F.: Neural fuzzy logic programming. IEEE Translations on Neural Networks 3(5) (1992)
Fu, L.M.: Knowledge-based connectionism from revising domain theories. IEEE Trans. Syst. Man. Cybern. 23, 173–182 (1993)
Towell, G.G., Shavlik, J.W.: Knowledge-based artificial neural networks. Artif. Intell., 70, 119–165 (1994)
Gallant, S.I.: Neural Network Learning and Expert Systems. MIT Press, Cambridge (1994)
Towell, G.G., Shavlik, J.W.: Extracting refined rules from knowledge-based neural networks. Mach. Learn. 13, 71–101 (1993)
Castro, J., Trillas, E.: The logic of neural networks. Mathware and Soft Computing 5, 23–27 (1998)
Hagan, M.T., Menhaj, M.: Training feedforward networks with marquardt algorithm. IEEE Transaction on Neural Networks 5(6), 989–993 (1999)
Hassibi, B., Stork, D.G., Wolf, G.J.: Optimal brain surgeon and general network pruning. In: IEEE International Conference on Neural Network, vol. 4(5), pp. 740–747 (1993)
Jipsen, P.: An overview of generalised basic logic algebra. Neural Network World 13(5), 491–500 (2003)
Gerla, B.: Functional representation of many-valued logics based on continuous t-norms. PhD thesis, University of Milano (2000)
Hájek, P.: Fuzzy logic from the logical point of view. In: Bartosek, M., Staudek, J., Wiedermann, J. (eds.) SOFSEM 1995. LNCS, vol. 1012. Springer, Heidelberg (1995)
Amato, P., Nola, A.D., Gerla, B.: Neural networks and rational łukasiewicz logic. IEEE Transaction on Neural Networks 5(6), 506–510 (2002)
Mundici, D.: A constructive proof of macnaughton’s theorem in intinite-valued logics. Journal of Symbolic Logic 59, 596–602 (1994)
Dubois, D., Prade, H.: Fundamentals of fuzzy sets. Kluwer, Dordrecht (2000)
Charalambous, C.: Conjugate gradient algorithm for efficient training of artificial neural networks. IEEE Proceedings 139(3), 301–310 (1992)
Battiti, R.: Frist- and second-order methods for learning between steepest descent and newton’s method. Neural Computation 4(2), 141–166 (1992)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2011 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Leandro, C., Pita, H., Monteiro, L. (2011). Symbolic Knowledge Extraction from Trained Neural Networks Governed by Łukasiewicz Logics. In: Madani, K., Correia, A.D., Rosa, A., Filipe, J. (eds) Computational Intelligence. IJCCI 2009. Studies in Computational Intelligence, vol 343. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-20206-3_3
Download citation
DOI: https://doi.org/10.1007/978-3-642-20206-3_3
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-20205-6
Online ISBN: 978-3-642-20206-3
eBook Packages: EngineeringEngineering (R0)