Abstract. In [18] we have shown how to construct a 3--layered recurrent neural network that computes the fixed point of the meaning function TP of a given propositional logic program P, which corresponds to the computation of the semantics of P. In this article we consider the first order case. We define a notion of approximation for interpretations and prove that there exists a 3--layered feed forward neural network that approximates the calculation of TP for a given first order acyclic logic program P with an injective level mapping arbitrarily well. Extending the feed forward network by recurrent connections we obtain a recurrent neural network whose iteration approximates the fixed point of TP. This result is proven by taking advantage of the fact that for acyclic logic programs the function TP is a contraction mapping on a complete metric space defined by the interpretations of the program. Mapping this space to the metric space IR with Euclidean distance, a real valued function fP can be defined which corresponds to TP and is continuous as well as a contraction. Consequently it can be approximated by an appropriately chosen class of feed forward neural networks.
|
1677
|
Foundations of Logic Programming
– Lloyd
- 1987
|
|
834
|
Multi-layer feed forward networks are universal aproximators
– Hornik, Stinchcombe, et al.
- 1989
|
|
805
|
Negation as failure
– Clark
- 1978
|
|
311
|
Recursive distributed representations
– Pollack
- 1990
|
|
272
|
On the approximate realization of continuous mappings by neural networks
– FUNAHASHI
- 1989
|
|
249
|
On the proper treatment of connectionism
– Smolensky
- 1988
|
|
237
|
On the declarative semantics of deductive databases and logic programs
– Przymusinski
- 1988
|
|
182
|
A survey and critique of techniques for extracting rules from trained arti cial neural networks
– Andrews, Diederich, et al.
- 1995
|
|
176
|
Contributions to the theory of logic programming
– Apt, Emden
- 1982
|
|
176
|
Learning and Extracting Finite State Automata with Second-Order Recurrent Neural Networks, Neural Computation
– Giles, Miller, et al.
- 1992
|
|
171
|
Extracting refined rules from knowledge-based neural networks
– Towell, Shavlik
- 1993
|
|
166
|
Formulas as programs
– Apt, Bezem
- 1999
|
|
129
|
SATCHMO: a theorem prover implemented in Prolog
– Manthey, Bry
|
|
103
|
The dynamics of discrete-time computation with applications to recurrent neural networks and finite state machine extraction
– Casey
- 1996
|
|
79
|
Holographic reduced representations
– Plate
- 1995
|
|
67
|
Metric methods: Three examples and a theorem
– FITTING
- 1994
|
|
67
|
Constructing deterministic finite-state automata in recurrent neural networks
– Omlin, Giles
- 1996
|
|
65
|
High-level Inferencing in a Connectionist Network
– Lange, Dyer
- 1989
|
|
64
|
Extraction of rules from discrete-time recurrent neural networks. Neural Networks
– Omlin, Giles
- 1996
|
|
57
|
General Topology
– Willard
- 1970
|
|
55
|
Distributed Representations and Nested Compositional Structure
– Plate
- 1994
|
|
48
|
Turing computability with neural nets
– Siegelmann, Sontag
- 1991
|
|
47
|
Labeling RAAM
– Sperduti
- 1994
|
|
41
|
Efficient simulation of finite automata by neural nets
– Alon, Dewdney, et al.
- 1991
|
|
38
|
Representation of finite state automata in recurrent radial basis function networks
– Frasconi, Gori, et al.
- 1996
|
|
36
|
Towards a massively parallel computational model for logic programming
– Hölldobler, Kalinke
- 1994
|
|
33
|
A Model-Guided Theorem Prover
– Scott
- 1993
|
|
31
|
From associations to systematic reasoning: A connectionist representation of rules, variables and dynamic bindings using temporal synchrony
– Shastri, Ajjanagadde
- 1993
|
|
22
|
Model generation theorem provers on a parallel inference machine
– Fujita, Hasegawa, et al.
- 1992
|
|
22
|
Induction of Finite-State Automata Using Second-Order Recurrent Networks
– Watrous, Kuhn
- 1992
|
|
20
|
Acyclic programs and the completeness of SLDNF-resolution
– Cavedon
- 1991
|
|
20
|
Symmetric neural networks and logic satisfiability
– Pinkas
- 1991
|
|
20
|
Topology and Iterates in Computational Logic
– Seda, Hitzler
- 1997
|
|
20
|
Strictly Level-Decreasing Logic Programs
– Seda, Hitzler
- 1998
|
|
19
|
Exploring the Computational Capabilities of Recurrent Neural Networks
– Kolen
- 1994
|
|
15
|
Automated inferencing and connectionist models
– Holldobler
- 1993
|
|
12
|
Computations in recurrent neural networks: From counters to iterated function systems
– Kalinke, Lehmann
- 1998
|
|
11
|
On the computational power of recurrent neural networks for structures
– Sperduti
- 1997
|
|
9
|
Using connectionist term representation for first–order deduction – a critical view
– Kalinke
- 1997
|
|
7
|
Epistemological challanges for connectionism
– McCarthy
- 1988
|
|
6
|
Generating Rational Models
– Bornscheuer
- 1996
|
|
6
|
Logic programming and inductive learning in artificial neural networks
– Garcez, Zaverucha, et al.
- 1997
|
|
5
|
Finite state automata that recurrent cascade-correlation cannot represent
– Kremer
- 1996
|
|
4
|
Recurrent Neural Networks to Approximate the Semantics of Acceptable Logic Programs
– Holldobler, Kalinke, et al.
- 1998
|
|
3
|
Expressing first-order logic in symmetric connectionist networks
– Pinkas
- 1991
|
|
2
|
On the approximate realization of continuous mappingsby neural networks
– Funahashi
- 1989
|
|
1
|
Yvonne Kalinke, and Antje Strohmaier. Massively Parallel Reasoning, volume II of Automated Deduction --- A Basis for Applications, chapter 11
– Bornscheuer, Holldobler
- 1998
|
|
1
|
Holldobler Steffen Holldobler has received his doctorate in science at the University of the Armed Forces
– Addison--Wesley
- 1970
|
|
