Abstract
The time when the connectionist and symbolic perspectives of Artificial Intelligence (AI) competed against each other is now over. The rivalry was due essentially to ignorance on the implications of the knowledge level, as introduced by Newell and Marr. Now it is generally accepted that they are different and complementary forms of modeling and operationalizing the inferences in terms of which a problem solving method (PSM) decomposes a task. All these tasks, methods, inferences, and formal operators belong to a broad library of reusable components for knowledge modeling. The final configuration of a problem solving method, with symbolic and connectionist components, is only dependent on the particular balance between data and knowledge available for the specific application under consideration. Various approaches have been explored for neurosymbolic integration. In this paper we propose a classification of these approaches (unified, hybrid and system level) and strongly support that the integration has to be made at the knowledge level and in the domain of the external observer (the “house” of models).
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
Clancey, W.J.: Situated Cognition. Camb. University Press, Cambridge (1997)
González, J.C., Velasco, J.R., Iglesias, C.A.: A distributed platform for symbolicconnectionist interoperation. In: Sun, R., Alexandre, F. (eds.) Connectionist-Symbolic Integration, pp. 175–193. Lawrence Erlbaum Asoc., London (1997)
Haykin, S.: Neural networks. A comprehensive foundation. Prentice-Hall, Englewood Cliffs (1999)
Hilario, M., et al.: Integration of model-based reasoning and neural processing. Report of MIX Project. CUI. University of Geneva (1999)
Hillario, M., Llamement, Y., Alexandre, F.: Neurosymbolic integration: Unified versus hybrid approaches. In: The European Symposium on Artificial Neural Networks, Brussels, Belgium (1995)
Kuncicky, D.C., Hruska, S.I., Lacher, R.C.: Hybrid systems: The equivalence of rulebased expert system and artificial neural network inference. International Journal of Expert System 4(3), 281–297 (1992)
Marr, D.: Vision. Freeman, New York (1982)
McCulloch, W.S., Pitts, W.: A logical calculus of the ideas immanent in nervous activity. Bulletin of Mathematical Biophysics 5, 115–133 (1943)
Medsker, L.R.: Hybrid Neural Network and Expert Systems. Kluwer Academic Publishers, Boston (1994)
Mira, J., Delgado, A.E.: Computación neuronal. En. Aspectos Básicos de la Inteligencia Artificial, Sanzy Torres. Madrid, pp. 485–576 (1995)
Mira, J., Delgado, A.E.: Where is knowledge in robotics? Some methodological issues on symbolic and connectionist perspectives of AI. In: Zhou, C., Maravall, D., Rua, D. (eds.) Autonomous Robotic Systems. Ch. I, pp. 3–34. Springer, Heidelberg (in press)
Mira, J., et al.: Aspectos básicos de la inteligencia artificial, Sanzy Torres. Madrid (1995)
Newell, A.: The knowledge level. AI Magazine, pp. 1–20 (summer 1981)
Sun, R., Bookman, L.: How do symbols and networks fit together. In: IA Magazine, pp. 20–23 (Summer 1993)
Towell, G.C., Shavlik, J.W.: Refining symbolic knowledge using neural networks. In: Michalshi, R., Tecuci, G. (eds.) Machine Learning, vol. IV, pp. 405–429. Morgan-Kaufmann Pub., C.A. (1994)
Towell, G.G.: Symbolic knowledge and neural networks: insertion refinement and extraction. Technical Report 1072, Univ. of Wisconsin-Madison, Computer Science Dept. (January 1992)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2003 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Mira, J., Delgado, A.E., Taboada, M.J. (2003). Neurosymbolic Integration: The Knowledge Level Approach. In: Moreno-Díaz, R., Pichler, F. (eds) Computer Aided Systems Theory - EUROCAST 2003. EUROCAST 2003. Lecture Notes in Computer Science, vol 2809. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-45210-2_42
Download citation
DOI: https://doi.org/10.1007/978-3-540-45210-2_42
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-20221-9
Online ISBN: 978-3-540-45210-2
eBook Packages: Springer Book Archive