Improving Out-of-Distribution Detection with Markov Logic Networks

TL;DR

This paper enhances out-of-distribution detection in deep learning by integrating Markov logic networks for probabilistic reasoning, improving performance and explainability across various datasets.

Contribution

It introduces a novel approach combining MLNs with existing OOD detectors and presents a new algorithm for learning logical constraints from data.

Findings

MLNs significantly improve OOD detection accuracy

The approach maintains computational efficiency

The logical constraint learning algorithm is effective

Abstract

Out-of-distribution (OOD) detection is essential for ensuring the reliability of deep learning models operating in open-world scenarios. Current OOD detectors mainly rely on statistical models to identify unusual patterns in the latent representations of a deep neural network. This work proposes to augment existing OOD detectors with probabilistic reasoning, utilizing Markov logic networks (MLNs). MLNs connect first-order logic with probabilistic reasoning to assign probabilities to inputs based on weighted logical constraints defined over human-understandable concepts, which offers improved explainability. Through extensive experiments on multiple datasets, we demonstrate that MLNs can significantly enhance the performance of a wide range of existing OOD detectors while maintaining computational efficiency. Furthermore, we introduce a simple algorithm for learning logical constraints for OOD detection from a dataset and showcase its effectiveness.