The work introduces a distributional view of visual mechanistic interpretability that casts the task as KL-minimal optimization and realizes it through a soft-constraint principle implemented with energy-guided diffusion posterior sampling on models such as DINOv3.
Title resolution pending
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
FaithTrace uses the directional derivative of class logits along text-induced directions in feature space as an influence score to produce and evaluate more faithful zero-shot textual explanations for image classifiers.
citing papers explorer
-
A Distributional View for Visual Mechanistic Interpretability: KL-Minimal Soft-Constraint Principle
The work introduces a distributional view of visual mechanistic interpretability that casts the task as KL-minimal optimization and realizes it through a soft-constraint principle implemented with energy-guided diffusion posterior sampling on models such as DINOv3.
-
Zero-Shot Faithful Textual Explanations via Directional-Derivative Influence on Predictions
FaithTrace uses the directional derivative of class logits along text-induced directions in feature space as an influence score to produce and evaluate more faithful zero-shot textual explanations for image classifiers.