Topic

Intelligent systems based on logical knowledge processing on the one hand, and on artificial neural networks (also called connectionist systems) on the other, differ substantially. They are both standard approaches to artificial intelligence and it would be very desirable to combine the robust neural networking machinery with symbolic knowledge representation and reasoning paradigms like logic programming in such a way that the strengths of either paradigm will be retained. Current research, however, fails by far to achieve this goal, in particular for non-propositional logics.

Contents: Overview

We will present past and present achievements in neural-symbolic integration. Starting from successful achievements for propositional logic, we will work towards discussing state-of-the-art research on the integration of first-order logic programming and connectionism, based on recent research publications by the organizors. The course shall emphasize the importance of neural-symbolic integration and stimulate research in this direction.

Course Material

Course material (slides) - final version:
Part I (pdf, 625KB)
Part II (zipped ps, 2.341KB)

Presenters Information

	M.Sc. Sebastian Bader is a Ph.D. student in Computer Science at the Technische Universität Dresden since 2004. He also received his M.Sc in Computational Logic in Dresden. His Dissertation is concerned with Neural-Symbolic Integration.
	Dr. Pascal Hitzler is assistant professor at the Institute for Applied Informatics and Formal Description Methods (AIFB) at the University of Karlsruhe in Germany, where he is involved in national and international projects on semantic web technologies, including KnowledgeWeb, SEKT, and SmartWeb. Before this, he was a postdoctoral researcher at the Artificial Intelligence Institute at Dresden University. He received a PhD in Mathematics from UCC Cork, Ireland in 2001, and a Diplom in Mathematics and Computer Science from the University of Tübingen, Germany, in 1998. His research record lists over 80 publications in such diverse areas as neural-symbolic integration, semantic web, knowledge representation and reasoning, lattice and domain theory, denotational semantics, and set-theoretic topology. He serves as a reviewer for international journals, conferences, and research project applications. He has also been an organizer of international enhancement programmes for highly skilled students in Mathematics and Computer Science, and has served as an editor for several books in this area.
	Prof. Dr. Steffen Hölldobler has received the title of a Dr. rer. nat. in 1988 from the University of the Armed Forces Munich. During a post-doctoral fellowship at the International Computer Science Institute in Berkeley, California, from 1989 to 1990 he was introduced to connectionism by Jerome Feldman and developed the core of his post-doctoral thesis on "Automated Inferencing and Connectionist Models". In 1993 he became Professor for "Knowledge Representation and Reasoning" at the Department of Computer Science of the Technische Universität Dresden. He is currently director of the International Center for Computational Logic and head of the European Masters Program in Computational Logic. Steffen Hölldobler has developed various connectionist models for propositional and first-order reasoning including a connectionist unification algorithm, a connectionist inference system for first-order Horn Logic (CHCL), a recursive propositional connectionist inference system for normal logic programs, and a recursive neural network for reflexive reasoning. He has also shown that the immediate consequence operator of certain classes of first-order logic programs can be approximated arbitrarily well by feedforward neural networks and, consequently, that the least fixed point of such programs can be approximated arbitrarily well by recurrent networks with a feedforward kernel.