Neural Logic Networks for Interpretable Classification
This work addresses the need for interpretable AI models in critical domains such as healthcare and industry, where understanding model decisions is essential, though it is incremental as it builds upon existing neural logic networks.
The paper tackles the problem of interpretability in neural networks by introducing Neural Logic Networks with NOT operations and biases, which learn logical mechanisms using AND, OR, and NOT operations for classification. The result is an improvement in state-of-the-art Boolean network discovery, enabling the learning of relevant, interpretable rules in domains like medical and industrial fields.
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.