arXiv cs.AI· Roshan Prakash Rane, Marco Simnacher, Manuel Pfeuffer, Marc-Andre Schulz, Nys Tjade Siegel, Maximilian Dreyer, Frederik Pahde, Wojciech Samek, Sonja Greven, Kerstin Ritter·· 2 天前AI 评分22
ICON 分解：多变量概念审计揭示神经网络依赖的捷径
ICON decomposition: Multivariate concept auditing reveals shortcuts that neural networks rely on
AI 导读
研究者提出 ICON（Independent Canonical cONcepts），将捷径审计重构为对中间表示方差的分解，量化每个概念在控制其他概念与结果后对某层方差的贡献，并报告未被解释的方差。在含真实标签的模拟数据中，ICON 比七个基线更准确地还原概念重要性；在皮肤癌与神经影像应用中，它能区分已学到的捷径与仅相关的概念，并经独立验证确认。
正文 · 原文
Abstract:Deep neural networks often exploit spurious associations, a failure known as shortcut learning. Before deployment, models should therefore be audited for reliance on concepts such as acquisition artifacts and patient demographics. Current methods, including linear probes and concept activation vectors, measure concept reliance by asking whether each concept is decodable from a layer. We show that this framing is misleading: decodability reflects not only concept reliance but also correlations among concepts and the outcome in the auditing dataset. We reframe shortcut auditing as a decomposition of the variance of intermediary representations with Independent Canonical cONcepts (ICON). ICON quantifies the share of a layer's variance explained by each concept conditional on all other concepts and the outcome, and also reports the variance left unexplained. Because ICON explanations are comparable across layers, their trajectories distinguish concepts that drive predictions from those that are merely encoded in the representations. In simulated data with ground truth, ICON recovers concept importance more accurately than seven baselines. In skin-cancer and neuroimaging applications, it separates learned shortcuts from merely correlated concepts, which we confirm with independent validation checks.
| Comments: | 44 pages, 12 figures, 3 tables. Includes Extended Data (7 figures, 2 tables). Code: this https URL |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (stat.ML) |
| ACM classes: | I.2.6; I.2.10; I.5.1; J.3 |
| Cite as: | arXiv:2608.26083 [cs.LG] |
| (or arXiv:2608.26083v4 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2608.26083 arXiv-issued DOI via DataCite |
Submission history
From: Roshan Prakash Rane [view email]
[v1]
Wed, 26 Aug 2026 17:47:49 UTC (3,663 KB)
[v2]
Mon, 31 Aug 2026 17:35:50 UTC (3,565 KB)
[v3]
Tue, 15 Sep 2026 17:58:01 UTC (3,570 KB)
[v4]
Thu, 15 Oct 2026 15:37:39 UTC (3,605 KB)
来源：arXiv cs.AI · arxiv.org