arXiv cs.AI· Nektarios Kalampalikis, Kavya Gupta, Georgi Vitanov, Isabel Valera·· 9 天前AI 评分22
面向合理概念瓶颈模型：CREAM 框架发布
Towards Reasonable Concept Bottleneck Models
AI 导读
研究者提出 Concept REAsoning Models（CREAM）框架，可在概念瓶颈模型（CBM）推理中显式编码概念-概念（C-C）与概念-任务（C→Y）关系，支持互斥、层级关联等任意类型关系及稀疏 C→Y 关系。CREAM 可选配正则化侧信道以补全不完整概念集，并引入 C→Y 无关的可解释性度量。实验显示其无需额外计算开销即可支持高效干预、避免概念泄漏，在概念缺失时达到黑盒级性能。
正文 · 原文
Abstract:We propose a novel, flexible, and efficient framework for designing Concept Bottleneck Models (CBMs) that enables practitioners to explicitly encode and extend their prior knowledge and beliefs about the concept-concept ($C-C$) and concept-task ($C \to Y$) relationships within the model's reasoning when making predictions. The resulting $\textbf{C}$oncept $\textbf{REA}$soning $\textbf{M}$odels (CREAMs) architecturally encode arbitrary types of $C-C$ relationships such as mutual exclusivity, hierarchical associations, and/or correlations, as well as potentially sparse $C \to Y$ relationships. Moreover, CREAM can optionally incorporate a regularized side-channel to complement the potentially {incomplete concept sets}, achieving competitive task performance while encouraging predictions to be concept-grounded. To evaluate CBMs in such settings, we introduce a $C \to Y$ agnostic metric that quantifies interpretability when predictions partially rely on the side-channel. In our experiments, we show that, without additional computational overhead, CREAM models support efficient interventions, can avoid concept leakage, and achieve black-box-level performance under missing concepts. We further analyze how an optional side-channel affects interpretability and intervenability. Importantly, the side-channel enables CBMs to remain effective even in scenarios where only a limited number of concepts are available.
| Comments: | 34 pages, 22 figures, Updated to the published version |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML) |
| Cite as: | arXiv:2506.05014 [cs.LG] |
| (or arXiv:2506.05014v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2506.05014 arXiv-issued DOI via DataCite |
|
| Journal reference: | Transactions on Machine Learning Research, 2026 |
Submission history
From: Nektarios Kalampalikis [view email]
[v1]
Thu, 5 Jun 2025 13:22:29 UTC (2,578 KB)
[v2]
Sat, 11 Apr 2026 16:01:20 UTC (3,323 KB)
[v3]
Thu, 8 Oct 2026 14:45:49 UTC (3,292 KB)
来源：arXiv cs.AI · arxiv.org