aiqbench analysis
This paper tackles a specific weakness in local permutation-based explainability: when a feature is perturbed, the replacement values often ignore its correlation with other inputs, producing unrealistic inputs the model was never trained on. LUCID addresses this by sampling replacements from the full conditional distribution, with theoretical links to existing importance measures and convergence and uncertainty guarantees. For ML practitioners, the value is more reliable per-prediction explanations in correlated data, though the approach depends on the quality of the conditional density estimate.
LUCID: Local Understanding of Covariate Importance with Conditional Density Estimation
arXiv:2610.35740v1 Announce Type: new Abstract: Computing local-level variable importance (VI) is critical for understanding machine learning models. Local permutation-based VI metrics offer a simple yet powerful approach, measuring how much a model's output changes when perturbing its inputs. However, existing perturbation strategies often ignore feature dependence or account for it only through restricted approximations. Perturbing a feature without fully respecting this dependence can generate unrealistic counterfactuals, thereby evaluating the model in regions unsupported by the observed data distribution. We address this gap by introducing a new model-agnostic local VI metric and estimator, Local Understanding of Covariate Importance with conditional Density estimation (LUCID). LUCID perturbs features of interest using plausible values drawn from the full conditional distribution given the remaining observed features. We establish theoretical connections between LUCID and existing VI measures, and provide an estimator with convergence guarantees and uncertainty quantification. Finally, we demonstrate LUCID's stable performance and practical utility across simulated and real-world examples.
Abstract:Computing local-level variable importance (VI) is critical for understanding machine learning models. Local permutation-based VI metrics offer a simple yet powerful approach, measuring how much a model's output changes when perturbing its inputs. However, existing perturbation strategies often ignore feature dependence or account for it only through restricted approximations. Perturbing a feature without fully respecting this dependence can generate unrealistic counterfactuals, thereby evaluating the model in regions unsupported by the observed data distribution. We address this gap by introducing a new model-agnostic local VI metric and estimator, Local Understanding of Covariate Importance with conditional Density estimation (LUCID). LUCID perturbs features of interest using plausible values drawn from the full conditional distribution given the remaining observed features. We establish theoretical connections between LUCID and existing VI measures, and provide an estimator with convergence guarantees and uncertainty quantification. Finally, we demonstrate LUCID's stable performance and practical utility across simulated and real-world examples.
Machine Learning (cs.LG); Machine Learning (stat.ML)
arXiv:2610.35740 [cs.LG]
(or arXiv:2610.35740v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.35740 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 · 247 words
Related coverage
ArtECulture: Benchmarking Culture-Conditioned Visual Emotion Understanding in Multimodal Large Language Models
arXiv · Aug 6, 2026
On the Optimality of Random Features under General Conditions: Misspecified Cases and Robust Learning
arXiv · Oct 14, 2026
From Wavefunction Regularity to Eigenvector Conditioning: Accuracy--Cost Trade-offs of Quantum Algorithms for Non-Hermitian Transcorrelated Hamiltonians
arXiv · Oct 1, 2026
Circuit priors in NN-bayes: more results
LessWrong · Oct 21, 2026