Using symbolic learning to improve knowledge-based neural networks

Authors:
Geoffrey G. Towell;Jude W. Shavlik
Affiliations:
Siemens Corporate Research, Princeton, NJ and University of Wisconsin, Madison, Wisconsin;University of Wisconsin, Madison, Wisconsin
Venue:
AAAI'92 Proceedings of the tenth national conference on Artificial intelligence
Year:
1992

Citing 5
Cited 2

Learning internal representations by error propagation

Parallel distributed processing: explorations in the microstructure of cognition, vol. 1
The cascade-correlation learning architecture

Advances in neural information processing systems 2
Training knowledge-based neural networks to recognize genes in DNA sequences

NIPS-3 Proceedings of the 1990 conference on Advances in neural information processing systems 3
Symbolic knowledge and neural networks: insertion, refinement and extraction

Symbolic knowledge and neural networks: insertion, refinement and extraction
Induction of Decision Trees

Machine Learning

GR2: a hybrid knowledge-based system using general rules

IJCAI'95 Proceedings of the 14th international joint conference on Artificial intelligence - Volume 1
Representations for a complex world: combining distributed and localist representations for learning and planning

Biomimetic Neural Learning for Intelligent Robots

Quantified Score

Hi-index	0.00

Visualization

Abstract

The previously-described KBANN system integrates existing knowledge into neural networks by defining the network topology and setting initial link weights. Standard neural learning techniques can then be used to train such networks, thereby refining the information upon which the network is based. However, standard neural learning techniques are reputed to have difficulty training networks with multiple layers of hidden units; KBANN commonly creates such networks. In addition, standard neural learning techniques ignore some of the information contained in the networks created by KBANN. This paper describes a symbolic inductive learning algorithm for training such networks that uses this previously-ignored information and which helps to address the problems of training "deep" networks. Empirical evidence shows that this method improves not only learning speed, but also the ability of networks to generalize correctly to testing examples.