Toward Faithful Explanations in Acoustic Anomaly Detection

2026-01-19

AI Summary

This study addresses the limited interpretability of deep learning models in audio anomaly detection, which often undermines user trust. By systematically comparing standard autoencoders with masked autoencoders and integrating attribution methods—namely SmoothGrad, Integrated Gradients, and Grad-CAM—the work evaluates the faithfulness and temporal precision of generated explanations. The authors propose a perturbation-based metric to assess explanation faithfulness and demonstrate that masked training significantly enhances explanation quality without compromising detection performance. Experiments on real-world industrial data show that masked autoencoders produce more faithful and temporally accurate explanations of anomalies, thereby substantially improving model trustworthiness.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Computer Vision: Interpretability, Explainability, and Transparency
• Philosophy and Ethics of AI: Accountability, Interpretability & Explainability

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Interpretability is essential for user trust in real-world anomaly detection applications. However, deep learning models, despite their strong performance, often lack transparency. In this work, we study the interpretability of autoencoder-based models for audio anomaly detection, by comparing a standard autoencoder (AE) with a mask autoencoder (MAE) in terms of detection performance and interpretability. We applied several attribution methods, including error maps, saliency maps, SmoothGrad, Integrated Gradients, GradSHAP, and Grad-CAM. Although MAE shows a slightly lower detection, it consistently provides more faithful and temporally precise explanations, suggesting a better alignment with true anomalies. To assess the relevance of the regions highlighted by the explanation method, we propose a perturbation-based faithfulness metric that replaces them with their reconstructions to simulate normal input. Our findings, based on experiments in a real industrial scenario, highlight the importance of incorporating interpretability into anomaly detection pipelines and show that masked training improves explanation quality without compromising performance.

Problem

• acoustic anomaly detection
• interpretability
• faithful explanations
• autoencoder
• mask autoencoder

Innovation

• mask autoencoder
• faithful explanation
• perturbation-based evaluation
• acoustic anomaly detection
• interpretability