通过频段扰动评估声学异常检测中AI解释的可信度
A Framework for Evaluating Faithfulness in Explainable AI for Machine Anomalous Sound Detection Using Frequency-Band Perturbation
- 用系统性频段移除法量化XAI解释与模型行为的一致性
- Occlusion方法最贴近真实敏感频段，梯度类方法常失准
- 为声学异常检测解释提供可复现的可信度评测标准
可解释人工智能（XAI）常用于声学异常检测（ASD）模型，以识别音频信号中影响异常判断的时间-频率区域。然而，多数音频解释依赖对显著性图的定性观察，无法确认这些归因是否真实反映模型所依赖的频谱线索。本文提出一种新的定量评估框架，通过系统性地移除特定频段，直接将归因相关性与模型行为关联，客观衡量XAI在机器声学分析中的忠实度。基于四种常用方法——Integrated Gradients、Occlusion、Grad-CAM和SmoothGrad——的实验表明，不同XAI技术可靠性差异显著：Occlusion与模型真实敏感度匹配最佳，而基于梯度的方法往往无法准确捕捉频谱依赖关系。该框架提供了可复现的音频解释基准评测方式，有助于提升基于频谱图的ASD系统解释的可信度。
原文摘要 · Abstract (English)
Explainable AI (XAI) is commonly applied to anomalous sound detection (ASD) models to identify which time-frequency regions of an audio signal contribute to an anomaly decision. However, most audio explanations rely on qualitative inspection of saliency maps, leaving open the question of whether these attributions accurately reflect the spectral cues the model uses. In this work, we introduce a new quantitative framework for evaluating XAI faithfulness in machine-sound analysis by directly linking attribution relevance to model behaviour through systematic frequency-band removal. This approach provides an objective measure of whether an XAI method for machine ASD correctly identifies frequency regions that influence an ASD model's predictions. By using four widely adopted methods, namely Integrated Gradients, Occlusion, Grad-CAM and SmoothGrad, we show that XAI techniques differ in reliability, with Occlusion demonstrating the strongest alignment with true model sensitivity and gradient-+based methods often failing to accurately capture spectral dependencies. The proposed framework offers a reproducible way to benchmark audio explanations and enables more trustworthy interpretation of spectrogram-based ASD systems.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。