analiza aiqbench
Reveal-IG introduces a novel attribution framework that shifts path-based explanations from raw input space to structured probe distributions, effectively resolving baseline artifacts and improving stability across image and tabular datasets. By progressively revealing information about input features, this method maintains the desirable completeness property while naturally accommodating multiscale probes and feature-wise uncertainty. The approach represents a significant methodological advance for researchers seeking more robust and sign-consistent feature importance scores in explainable AI.
Attribution via Distributional Paths for Information Revelation
arXiv:2606.03885v1 Announce Type: new Abstract: Feature attribution methods explain predictions by assigning importance scores to input features. Path-based methods such as Integrated Gradients are especially appealing because they satisfy \textit{completeness}: attributions sum to the change in model output between a reference state and the input. Yet most path methods define this trajectory in input space, explaining a model through pointwise perturbed inputs along a chosen path. An input-space path integrates the model's raw response at each point it passes through, with no control over the resolution at which a feature is queried; the early, baseline-adjacent part of the trajectory contributes to the explanation on equal footing with the input itself. Here, we lift path attribution from input space to a space of structured probe distributions around the example of interest, and call our method Reveal-IG. Rather than traversing raw input values, Reveal-IG progressively reveals information about the input and attributes changes in the model's expected output along this distributional path. The result is a path-attribution framework that retains completeness with respect to the expected model response, and naturally accommodates multiscale image probes and feature-wise uncertainty in tabular data. Synthetic diagnostics show that Reveal-IG avoids path artifacts that affect input-space methods, and across ImageNet classification and tabular regression it produces stable, signed attributions -- leading on metrics that use attribution sign while remaining competitive on the rest.
Abstract:Feature attribution methods explain predictions by assigning importance scores to input features. Path-based methods such as Integrated Gradients are especially appealing because they satisfy \textit{completeness}: attributions sum to the change in model output between a reference state and the input. Yet most path methods define this trajectory in input space, explaining a model through pointwise perturbed inputs along a chosen path. An input-space path integrates the model's raw response at each point it passes through, with no control over the resolution at which a feature is queried; the early, baseline-adjacent part of the trajectory contributes to the explanation on equal footing with the input itself. Here, we lift path attribution from input space to a space of structured probe distributions around the example of interest, and call our method Reveal-IG. Rather than traversing raw input values, Reveal-IG progressively reveals information about the input and attributes changes in the model's expected output along this distributional path. The result is a path-attribution framework that retains completeness with respect to the expected model response, and naturally accommodates multiscale image probes and feature-wise uncertainty in tabular data. Synthetic diagnostics show that Reveal-IG avoids path artifacts that affect input-space methods, and across ImageNet classification and tabular regression it produces stable, signed attributions -- leading on metrics that use attribution sign while remaining competitive on the rest.
Machine Learning (cs.LG)
arXiv:2606.03885 [cs.LG]
(or arXiv:2606.03885v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2606.03885 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.
Text integral sub licența deschisă a publicației · 315 cuvinte
Subiecte conexe
QuanText: Protecting Dataset-Level Secrets in Textual Data Sharing
arXiv · 17 sept. 2026
An Interpretable Deep Learning Framework for Material Perception and Classification from Multisensory Tactile Data
arXiv · 25 aug. 2026
A Unified Framework for Uncertainty-Aware Explainable Artificial Intelligence: A Case Study in Power Quality Disturbance Classification
arXiv · 29 iul. 2026
Statistical attribute alignment for black-box generative AI via output post-processing
arXiv · 28 sept. 2026