arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2604.02532cs.CVcs.AIcs.LG
收藏

特征归因稳定性套件:后验归因的稳定性如何?

Feature Attribution Stability Suite: How Stable Are Post-Hoc Attributions?

The George Washington University(乔治华盛顿大学)

Kamalasankari Subramaniakuppusamy, Jugal Gajjar

AI总结:

本文提出FASS基准,通过预测不变过滤分解稳定性为结构相似度、排名相关性和top-k Jaccard重叠,评估四种归因方法在不同扰动下的稳定性,发现几何扰动比光度扰动更具不稳定性。

摘要:

后验特征归因方法广泛应用于安全关键视觉系统,但其在现实输入扰动下的稳定性仍缺乏充分研究。现有指标主要在加性噪声下评估解释,稳定性归一化为单一标量,并未考虑预测保持,混淆了解释脆弱性与模型敏感性。本文引入特征归因稳定性套件(FASS),一个强制预测不变过滤的基准,将稳定性分解为三个互补指标:结构相似度、排名相关性和top-k Jaccard重叠,并在几何、光度和压缩扰动下评估。评估四种归因方法(集成梯度、梯度SHAP、Grad-CAM、LIME)在四种架构和三个数据集(ImageNet-1K、MS COCO、CIFAR-10)上的表现,发现稳定性估计依赖于扰动家族和预测不变过滤。几何扰动暴露出比光度变化更大的归因不稳定性,且未考虑预测保持时,评估的99%配对涉及预测变化。在受控评估中,观察到一致的方法级趋势,Grad-CAM在所有数据集上实现最高稳定性。

英文摘要:

Post-hoc feature attribution methods are widely deployed in safety-critical vision systems, yet their stability under realistic input perturbations remains poorly characterized. Existing metrics evaluate explanations primarily under additive noise, collapse stability to a single scalar, and fail to condition on prediction preservation, conflating explanation fragility with model sensitivity. We introduce the Feature Attribution Stability Suite (FASS), a benchmark that enforces prediction-invariance filtering, decomposes stability into three complementary metrics: structural similarity, rank correlation, and top-k Jaccard overlap-and evaluates across geometric, photometric, and compression perturbations. Evaluating four attribution methods (Integrated Gradients, GradientSHAP, Grad-CAM, LIME) across four architectures and three datasets-ImageNet-1K, MS COCO, and CIFAR-10, FASS shows that stability estimates depend critically on perturbation family and prediction-invariance filtering. Geometric perturbations expose substantially greater attribution instability than photometric changes, and without conditioning on prediction preservation, up to 99% of evaluated pairs involve changed predictions. Under this controlled evaluation, we observe consistent method-level trends, with Grad-CAM achieving the highest stability across datasets.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑