Incorporating Prior Knowledge in the Form of Production Rules into Neural Networks Using Boolean-Like Neurons

Authors:
Songhe Zhao;Tharam S. Dillon
Affiliations:
School of Computer Science & Computer Engineering, La Trobe University, Bundoora, VIC 3083, Australia. E-mail: zhao@latcsl.cs.latrobe.edu.au;School of Computer Science & Computer Engineering, La Trobe University, Bundoora, VIC 3083, Australia. E-mail: zhao@latcsl.cs.latrobe.edu.au
Venue:
Applied Intelligence
Year:
1997

Citing 3
Cited 0

A Method for Learning From Hints

Advances in Neural Information Processing Systems 5, [NIPS Conference]
Network Structuring and Training Using Rule-Based Knowledge

Advances in Neural Information Processing Systems 5, [NIPS Conference]
Using Prior Knowledge in a {NNPDA} to Learn Context-Free Languages

Advances in Neural Information Processing Systems 5, [NIPS Conference]

Quantified Score

Hi-index	0.00

Visualization

Abstract

At present, nearly all neural networks are formulated bylearning only from examples or patterns. For a real-word problem,some forms of prior knowledge in a non-example form always exist.Incorporation of prior knowledge will benefit the formulation ofneural networks. Prior knowledge could be in several forms.Production rule is one form in which the prior knowledge isfrequently represented. This paper proposes an approach toincorporate production rules into neural networks. A newly definedneuron architecture, Boolean-like neuron, is proposed. With thisBoolean-like neuron, production rules can be encoded into the neuralnetwork during the network initialization period. Experiments aredescribed in this paper. The results show that the incorporation ofthis prior knowledge can not only increase the training speed, butalso the explainability of the neural networks.