Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhou, Guancheng, Luo, Yisi, He, Zhengfu, Jin, Zhenyu, Ge, Xuyang, Shu, Wentao, Meng, Deyu, Qiu, Xipeng
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2605.17504
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866913137018535936
author Zhou, Guancheng
Luo, Yisi
He, Zhengfu
Jin, Zhenyu
Ge, Xuyang
Shu, Wentao
Meng, Deyu
Qiu, Xipeng
author_facet Zhou, Guancheng
Luo, Yisi
He, Zhengfu
Jin, Zhenyu
Ge, Xuyang
Shu, Wentao
Meng, Deyu
Qiu, Xipeng
contents Most current paradigms in visual mechanistic interpretability (MI) remain confined to interpreting internal units of the vision model via heuristic methods (e.g., top-$K$ activation retrieval or optimization with regularization). In this work, we establish a theoretical distributional view for visual MI, which models the influence of a feature activation on the natural image distribution, thereby formulating a Kullback-Leibler (KL)-minimal optimization problem to model the MI task. Under this framework, statistical biases are identified within previous MI paradigms, which reveal that they may either be perceptually uninterpretable to humans (i.e., deviate from the natural image distribution), or mechanistically unfaithful to the vision models (i.e., unable to activate model features). To resolve the biases under the distributional view, we propose a model with a KL-minimal soft-constraint principle for visual MI that theoretically balances interpretability and faithfulness. We realize this principle via energy-guided diffusion posterior sampling. Extensive experiments validate the theoretical soundness of the proposed distributional view and demonstrate the practical effectiveness of our paradigm on the DINOv3 vision model.
format Preprint
id arxiv_https___arxiv_org_abs_2605_17504
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle A Distributional View for Visual Mechanistic Interpretability: KL-Minimal Soft-Constraint Principle
Zhou, Guancheng
Luo, Yisi
He, Zhengfu
Jin, Zhenyu
Ge, Xuyang
Shu, Wentao
Meng, Deyu
Qiu, Xipeng
Computer Vision and Pattern Recognition
Artificial Intelligence
Most current paradigms in visual mechanistic interpretability (MI) remain confined to interpreting internal units of the vision model via heuristic methods (e.g., top-$K$ activation retrieval or optimization with regularization). In this work, we establish a theoretical distributional view for visual MI, which models the influence of a feature activation on the natural image distribution, thereby formulating a Kullback-Leibler (KL)-minimal optimization problem to model the MI task. Under this framework, statistical biases are identified within previous MI paradigms, which reveal that they may either be perceptually uninterpretable to humans (i.e., deviate from the natural image distribution), or mechanistically unfaithful to the vision models (i.e., unable to activate model features). To resolve the biases under the distributional view, we propose a model with a KL-minimal soft-constraint principle for visual MI that theoretically balances interpretability and faithfulness. We realize this principle via energy-guided diffusion posterior sampling. Extensive experiments validate the theoretical soundness of the proposed distributional view and demonstrate the practical effectiveness of our paradigm on the DINOv3 vision model.
title A Distributional View for Visual Mechanistic Interpretability: KL-Minimal Soft-Constraint Principle
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2605.17504