Neural-Symbolic Integration for Interactive Learning and Conceptual Grounding

TL;DR

This paper introduces a neural-symbolic integration framework that enables interactive learning and conceptual explanation of neural models through symbolic logic queries and user feedback, enhancing interpretability and model refinement.

Contribution

It presents a novel method combining neural networks with symbolic logic for interactive learning and explanation, demonstrated with the Logic Tensor Network framework.

Findings

Enables querying neural models with symbolic logic language

Allows user interaction to refine models using logic-based constraints

Applied successfully to CNN with Concept Activation Vectors

Abstract

We propose neural-symbolic integration for abstract concept explanation and interactive learning. Neural-symbolic integration and explanation allow users and domain-experts to learn about the data-driven decision making process of large neural models. The models are queried using a symbolic logic language. Interaction with the user then confirms or rejects a revision of the neural model using logic-based constraints that can be distilled into the model architecture. The approach is illustrated using the Logic Tensor Network framework alongside Concept Activation Vectors and applied to a Convolutional Neural Network.