评估视觉模型解释的稳定性，发现几何扰动影响更大。
Feature Attribution Stability Suite: How Stable Are Post-Hoc Attributions?
- 提出FASS基准，控制预测不变性，分解稳定性为三维度指标
- 几何扰动下解释不稳定性显著高于光照变化，99%扰动对预测有影响
- Grad-CAM在多数据集上最稳定，适合高可靠性场景
后处理特征归因方法广泛应用于安全关键视觉系统，但其在真实输入扰动下的稳定性尚未充分刻画。现有度量主要针对加性噪声，将稳定性简化为单一标量，且未考虑预测保持，混淆了解释脆弱性与模型敏感性。我们提出特征归因稳定性套件（FASS），引入预测不变性过滤，将稳定性分解为结构相似性、秩相关性和top-k Jaccard重叠三个互补指标，并在几何、光度和压缩扰动下进行评估。在四个归因方法（Integrated Gradients、GradientSHAP、Grad-CAM、LIME）、四种架构及三个数据集（ImageNet-1K、MS COCO、CIFAR-10）上测试表明，稳定性估计高度依赖扰动类型与预测不变性条件。几何扰动导致的归因不稳定性远高于光度变化，且在未施加预测不变性约束时，高达99%的样本对预测结果产生改变。在受控评估下，观测到方法级一致趋势，其中Grad-CAM在所有数据集上表现最优。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。