Modular design patterns for neural-symbolic integration: refinement and combination

TL;DR

This paper formalizes neural-symbolic design patterns, enabling modular combination and refinement of patterns, with implementation in the Hets tool set for validation and composition.

Contribution

It introduces formal definitions for pattern refinement and modular combination, advancing neural-symbolic integration methods.

Findings

Patterns and refinements can be checked for well-formedness.

Modular combinations of patterns can be computed automatically.

Implementation in Hets facilitates practical application.

Abstract

We formalise some aspects of the neural-symbol design patterns of van Bekkum et al., such that we can formally define notions of refinement of patterns, as well as modular combination of larger patterns from smaller building blocks. These formal notions are being implemented in the heterogeneous tool set (Hets), such that patterns and refinements can be checked for well-formedness, and combinations can be computed.