Publication
Neural-Symbolic Predicate Invention: Learning Relational Concepts from Visual Scenes
Jingyuan Sha; Hikaru Shindo; Kristian Kersting; Devendra Singh Dhami
In: Artur S. d'Avila Garcez; Tarek R. Besold; Marco Gori; Ernesto Jiménez-Ruiz (Hrsg.). Proceedings of the 17th International Workshop on Neural-Symbolic Learning and Reasoning. International Workshop on Neural-Symbolic Learning and Reasoning (NeSy), July 3-5, Italy, Pages 103-117, CEUR Workshop Proceedings, Vol. 3432, CEUR-WS.org, 2023.
Abstract
The predicates used for Inductive Logic Programming (ILP) systems are usually elusive and need to be hand-crafted in advance, which limits the generalization of the system when learning new rules without sufficient background knowledge. Predicate Invention (PI) for ILP is the problem of discovering new concepts that describe hidden relationships in the domain. PI can mitigate the generalization problem for ILP by inferring new concepts, giving the system a better vocabulary to compose logic ruless. Although there are several PI approaches for symbolic ILP systems, PI for NeSy ILP systems that can handle visual input to learn logical rules using differentiable reasoning is relatively unaddressed. To this end, we propose a neural-symbolic approach, NeSy-π, to invent predicates from visual scenes for NeSy ILP systems based on clustering and extension of relational concepts.(π denotes the abbrivation of Predicate