An approach to supervised learning of three valued Lukasiewicz logic in Hölldobler's core method

Harder, F. & Besold, T. R. (2017). An approach to supervised learning of three valued Lukasiewicz logic in Hölldobler's core method. CEUR Workshop Proceedings, 1895, pp. 24-37.

Abstract

The core method [6] provides a way of translating logic programs into a multilayer perceptron computing least models of the programs. In [7] , a variant of the core method for three valued Lukasiewicz logic and its applicability to cognitive modelling were introduced. Building on these results, the present paper provides a modified core suitable for supervised learning, implements and executes supervised learning with the backpropagation algorithm and, finally, constructs a rule extraction method in order to close the neural-symbolic cycle.

Publication Type:	Article
Additional Information:	Copyright, 2017, the authors.
Publisher Keywords:	Neural networks, logic programs, neural-symbolic integration
Departments:	School of Science & Technology > Computer Science