音频分类中的神经表征与符号语义桥接研究
Bridging Neural Representations and Symbolic Semantics in Audio Classification
AI 导读
一项 arXiv 研究提出模型无关的神经符号框架，通过符号概念归纳分析音频分类器最后隐藏层的 32 个神经元，检验其能否对应人类可理解的声学概念。该框架在 GTZAN 音乐流派分类和 IRMAS 乐器分类两个任务上评估，结果显示多个隐藏神经元可关联到有意义的概念，多数确认的神经元-概念对经 Mann-Whitney U 检验获得统计支持。
正文 · 原文
Abstract:Deep audio classifiers can achieve strong predictive performance, but their hidden representations remain difficult to interpret. Existing audio explainability methods often focus on input-level attribution, such as waveform segments, spectrogram regions, or acoustic features, and provide limited insight into what hidden neurons encode. Interpreting hidden neurons at the concept level can provide a more meaningful view of what an audio model learns internally. This paper investigates whether hidden neurons in audio classifiers can be associated with human-understandable symbolic acoustic concepts. A model-agnostic neurosymbolic framework is presented for analyzing the final hidden layer of audio classifiers, containing 32 neurons, through symbolic concept induction. The framework is evaluated on two audio classification settings: music genre classification using GTZAN dataset and musical instrument classification using IRMAS dataset. Results show that several hidden neurons can be associated with meaningful induced concepts, and most confirmed neuron-concept pairs are statistically supported using the Mann-Whitney U test. These findings suggest that symbolic concept induction is a promising direction for explaining latent audio representations beyond input-level attribution.
| Subjects: | Sound (cs.SD); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.40414 [cs.SD] |
| (or arXiv:2610.40414v1 [cs.SD] for this version) | |
| https://doi.org/10.48550/arXiv.2610.40414 arXiv-issued DOI via DataCite |
Submission history
From: Avishek Das [view email]
[v1]
Thu, 29 Oct 2026 16:42:13 UTC (1,111 KB)
来源：arXiv cs.AI · arxiv.org