Abstract
In this work we determine hyper-plane equations from three MLP models. The first one is the standard MLP model, the second one is called OMLP (oblique MLP) and the last one is called IMLP (Interpretable MLP). From OMLP and IMLP, hyper-plane equations are determined easily, whereas for MLP we just give a sufficient condition for the detection of potential hyper-plane discriminators.
Our goal is to justify MLP classification responses in terms of symbolic rules. For this, we use a standard MLP network for classification and an IMLP network for justification of MLP responses. The system consists in the training of the IMLP network with MLP responses and in the extraction of symbolic rules from IMLP. The approach is sufficiently general to work even when input variables are continuous. Moreover, the justification provided by IMLP is accurate because if MLP and IMLP responses are contradictory with respect to a new unknown example, IMLP is retrained with the addition of the new example until the system becomes coherent. Finally, we show results given by a medical diagnosis application with continuous input variables.
Preview
Unable to display preview. Download preview PDF.
References
White H.: Connectionist non-parametric regression. multi-layer feedforward networks can learn arbitrary mappings. Neural Networks. 3 (1990) 535–551.
Breimann L., Friedmann J.H., Olshen R.A., Stone J.: Classification and Regression Trees. Wadsworth and Brooks, Monterey, Calif (1984).
Quinlan J.R.: C4.5: Programs for Machine Learning. Morgan Kaufmann (1993).
Mooney R., Shavlik J., Towell G., Gove A.: An experimental Comparison of Symbolic and Connectionist Learning Algorithms. Proc. IJCAI-89 Morgan Kaufmann Los Altos. (1989) 775–780.
Atlas L., Cole R., Connor J., El-Sharkawi M., Marks R.J., Muthusumi Y., Barnard E.: Performance Comparison Between Backpropagation Networks and Classification Trees on Three Real-World Applications. NIPS 2, Morgan Kaufmann, San Mateo, CA. (1990) 622–629.
Tsoi A.C., Pearson R.A.: Comparison of Three Classification Techniques, CART, C4.5 and MLP. NIPS 3, Morgan Kaufmann, San Mateo CA. (1991) 963–969.
Mitchell T.M., Thsun S.B.: Explanation Based Learning. A comparison of symbolic and connectionist Learning Algorithms. Proc. 10th Int. Conf. on Machine Learning, Morgan Kaufmann San Mateo CA. (1993) 197–204.
Feng G., Sutherland A., King R., Muggleton S., Henery R.: Comparison of Machine Learning Classifiers to Statistics and Neural Networks. Proc. 4th Int. Workshop on Artificial Intelligence and Statistics, Florida (1993).
Quinlan J.R.: Comparing Connectionist and Symbolic Learning Methods. Hanson et al, (1994) 445–456.
Towell G.G., Shawlik J.W.: Extracting Refined Rules from Knowledge-Based Neural Networks. Machine Learning. 13 (1993) (1).
Setiono R., Liu H.: Understanding Neural Networks via Rule Extraction. Proc. IJCAI. 1 (1995) 480–485.
Gorman R.P., Sejnowski T.J.: Analysis of Hidden Units in a Layered Network Trained to Classify Sonar Targets. Neural Networks. 1 (1988) 75–88.
Andrews R., Geva S.: Extracting Rules From a Constrained Error Backpropagation Network. Proc. of the 5th Australian Conference on Neural Networks, Brisbane (1994).
Bologna G., Pellegrini C.: Extraction de Règles d'un réseau PMC à Valeurs d'Entrée Continues utilisant des Unités Cachées à Seuil dans la Couche Cachée. Proc. des Huitièmes journées Neurosciences et Sciences de l'Ingénieur, (1996) 199–204.
Bologna G.: Rule Extraction from the IMLP Neural Network: a Comparative Study. Proc. of the NIPS-96 Workshop of Rule Extraction from Trained Artificial Neural Network, (1996) 13–19.
Huang H.H., Zhang C., Lee S.: Implementation and Comparison of Neural Network Learning Paradigms: Back Propagation, Simulated Annealing and Tabu Search. Artificial Neural Networks in Engineering 1991: ASME Press, New York (1991) 95–100.
Amendolia S.R., Bertolucci E., Biadi O., Bottigli U., Caravelli P., Fantacci M.E., Fidecaro E., Mariani M., Messineo A., Rosso V., Stefanini A.: Neural Network Expert System for Screening Coronary Heart Disease. Physica Medica 1993: 9 (1); 13–17.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1997 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Bologna, G., Pellegrini, C. (1997). Accurate decomposition of standard MLP classification responses into symbolic rules. In: Mira, J., Moreno-Díaz, R., Cabestany, J. (eds) Biological and Artificial Computation: From Neuroscience to Technology. IWANN 1997. Lecture Notes in Computer Science, vol 1240. Springer, Berlin, Heidelberg. https://doi.org/10.1007/BFb0032521
Download citation
DOI: https://doi.org/10.1007/BFb0032521
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-63047-0
Online ISBN: 978-3-540-69074-0
eBook Packages: Springer Book Archive