Neural-Symbolic Integration with Evolvable Policies

2026-01-08

arXiv.org

AI Summary

This work proposes a novel evolutionary approach to neural-symbolic integration that overcomes the limitations of existing systems, which typically rely on predefined or differentiable symbolic policies and thus struggle in settings lacking expert knowledge or when policies are non-differentiable. The method treats the neural-symbolic system as an evolvable individual, simultaneously learning non-differentiable symbolic policies and neural network weights from an initial state of an empty policy and random weights. By integrating the NEUROLOG architecture, Valiant’s evolvability framework, Machine Coaching semantics, and abduction-driven neural training, the system evolves through mutation and fitness-based selection to approximate the target policy. Experimental results demonstrate that the approach can effectively learn complex symbolic strategies without any prior knowledge, achieving near-perfect median accuracy (~100%) and significantly extending the applicability of neural-symbolic systems to scenarios without expert guidance.

Technology Category

• Machine Learning: Neuro-Symbolic Learning
• Cognitive Modeling & Cognitive Systems: Symbolic Representations
• Multiagent Systems: Adversarial Agents

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Social networks and social learning
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Neural-Symbolic (NeSy) Artificial Intelligence has emerged as a promising approach for combining the learning capabilities of neural networks with the interpretable reasoning of symbolic systems. However, existing NeSy frameworks typically require either predefined symbolic policies or policies that are differentiable, limiting their applicability when domain expertise is unavailable or when policies are inherently non-differentiable. We propose a framework that addresses this limitation by enabling the concurrent learning of both non-differentiable symbolic policies and neural network weights through an evolutionary process. Our approach casts NeSy systems as organisms in a population that evolve through mutations (both symbolic rule additions and neural weight changes), with fitness-based selection guiding convergence toward hidden target policies. The framework extends the NEUROLOG architecture to make symbolic policies trainable, adapts Valiant's Evolvability framework to the NeSy context, and employs Machine Coaching semantics for mutable symbolic representations. Neural networks are trained through abductive reasoning from the symbolic component, eliminating differentiability requirements. Through extensive experimentation, we demonstrate that NeSy systems starting with empty policies and random neural weights can successfully approximate hidden non-differentiable target policies, achieving median correct performance approaching 100%. This work represents a step toward enabling NeSy research in domains where the acquisition of symbolic knowledge from experts is challenging or infeasible.

Problem

• Neural-Symbolic Integration
• non-differentiable policies
• symbolic knowledge acquisition
• evolvable policies
• interpretable reasoning

Innovation

• Neural-Symbolic Integration
• Evolvable Policies
• Non-differentiable Symbolic Reasoning
• Abductive Learning
• Machine Coaching