Neural Logic Networks for Interpretable Classification

TL;DR

This paper introduces Neural Logic Networks with enhanced interpretability through logical operations and probabilistic modeling, enabling the extraction of meaningful rules for classification tasks, especially in sensitive fields.

Contribution

It generalizes Neural Logic Networks with NOT operations and biases, proposes a new rule structure and learning algorithm, improving Boolean network discovery and interpretability.

Findings

Outperforms state-of-the-art Boolean network discovery methods.

Learns relevant, interpretable rules in tabular classification.

Effective in medical and industrial applications.

Abstract

Traditional neural networks have an impressive classification performance, but what they learn cannot be inspected, verified or extracted. Neural Logic Networks on the other hand have an interpretable structure that enables them to learn a logical mechanism relating the inputs and outputs with AND and OR operations. We generalize these networks with NOT operations and biases that take into account unobserved data and develop a rigorous logical and probabilistic modeling in terms of concept combinations to motivate their use. We also propose a novel factorized IF-THEN rule structure for the model as well as a modified learning algorithm. Our method improves the state-of-the-art in Boolean networks discovery and is able to learn relevant, interpretable rules in tabular classification, notably on examples from the medical and industrial fields where interpretability has tangible value.