深度神经网络机制可解释性：方法、框架与展望
Mechanistic Interpretability of Deep Neural Networks: Methods, Frameworks, and Perspectives
AI 导读
Lorenzo Cirillo 在 arXiv 发表综述，系统梳理深度神经网络机制可解释性研究的进展，聚焦通过识别对特定行为有因果作用的组件来揭示网络内部机制。文章指出内部激活是跨层变换的结构化信号，单个神经元常编码多个相互关联的概念，因而难以解释。内容涵盖方法、框架与未来展望，涉及图像、音频和语言处理等信号领域。
正文 · 原文
Abstract:Deep neural networks underpin fundamental tasks in a wide range of signal domains, such as image, audio, and language processing. However, their internal mechanisms are notably difficult to interpret. Internal activations are structured signals transformed across layers that may encode a multitude of interdependent relationships, as individual neurons frequently represent multiple underlying concepts. Mechanistic interpretability aims to uncover the internal mechanisms of neural networks by identifying components whose activations are causally responsible for specific behaviors. In such a context, this article provides an overview of advances in mechanistic interpretability for the analysis of causal inference in deep neural networks.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2611.13868 [cs.LG] |
| (or arXiv:2611.13868v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2611.13868 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Lorenzo Cirillo [view email]
[v1]
Tue, 10 Nov 2026 10:29:14 UTC (4,659 KB)
来源：arXiv cs.AI · arxiv.org