aiqbench analysis
This study clarifies that interpretability in Concept Bottleneck Models does not automatically guarantee robustness, revealing instead a trade-off where sensitivity is redistributed rather than eliminated. By distinguishing between geometric and semantic perturbations, the research demonstrates that these models shift vulnerability across different perturbation spaces depending on task structure and vocabulary size. The findings underscore that interpretable intermediate representations serve distinct objectives from robustness, requiring careful evaluation frameworks to understand their specific failure modes.
Interpretable but Fragile? Robustness of Concept Bottlenecks under Geometric-Semantic Perturbations
arXiv:2609.38625v1 Announce Type: cross Abstract: Concept Bottleneck Models (CBMs) are designed to provide interpretable intermediate representations, yet how such bottlenecks affect robustness remains unclear, with existing studies reporting mixed and sometimes contradictory findings. We argue that these discrepancies arise from conflating different robustness notions and perturbation regimes, rather than from fundamental disagreements about CBMs themselves. To disentangle these factors, we introduce a generator-based evaluation framework that enables controlled comparisons between standard classifiers and CBMs under two distinct perturbation types: continuous geometric perturbations in latent space and discrete semantic interventions in concept space. Within this framework, we evaluate robustness both empirically, via prediction and concept-level sensitivity metrics, and certifiably, using randomized smoothing in latent and concept spaces. Across experiments, we reconcile previously conflicting findings by clarifying when, and in what sense, concept bottlenecks do or do not improve robustness. By further analyzing robustness under varying task conditions, including class semantic similarity and concept vocabulary size, we show that interpretability does not inherently confer robustness. Instead, concept bottlenecks shift where and how sensitivity manifests, revealing a nuanced interpretability robustness trade off that depends critically on the perturbation regime and task structure. Together, our results show that interpretability and robustness are distinct objectives: interpretable intermediate representations do not uniformly improve robustness, but instead redistribute sensitivity across perturbation spaces and model families.
Title:Interpretable but Fragile? Robustness of Concept Bottlenecks under Geometric-Semantic Perturbations
Abstract:Concept Bottleneck Models (CBMs) are designed to provide interpretable intermediate representations, yet how such bottlenecks affect robustness remains unclear, with existing studies reporting mixed and sometimes contradictory findings. We argue that these discrepancies arise from conflating different robustness notions and perturbation regimes, rather than from fundamental disagreements about CBMs themselves. To disentangle these factors, we introduce a generator-based evaluation framework that enables controlled comparisons between standard classifiers and CBMs under two distinct perturbation types: continuous geometric perturbations in latent space and discrete semantic interventions in concept space. Within this framework, we evaluate robustness both empirically, via prediction and concept-level sensitivity metrics, and certifiably, using randomized smoothing in latent and concept spaces. Across experiments, we reconcile previously conflicting findings by clarifying when, and in what sense, concept bottlenecks do or do not improve robustness. By further analyzing robustness under varying task conditions, including class semantic similarity and concept vocabulary size, we show that interpretability does not inherently confer robustness. Instead, concept bottlenecks shift where and how sensitivity manifests, revealing a nuanced interpretability robustness trade off that depends critically on the perturbation regime and task structure. Together, our results show that interpretability and robustness are distinct objectives: interpretable intermediate representations do not uniformly improve robustness, but instead redistribute sensitivity across perturbation spaces and model families.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2609.38625 [cs.LG]
(or arXiv:2609.38625v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.38625 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 320 words
Related coverage
Concept-based Visual Counterfactual Explanations with Diffusion Models
arXiv · Jul 28, 2026
A Geometric Unification of Concept Learning with Concept Cones
arXiv · Jun 9, 2026
[AINews] not much happened today
Latent Space · Sep 18, 2026
Disentangling Hallucinations: Orthogonal Semantic Projection for Robust Interpretability
arXiv · Jun 16, 2026