Skip to content
conftrace_
2026 ICML ICML 2026

Position: Explanation Stability Is a Property of the Model–Method Pair, Not the Model

This position paper argues that explanation stability claims are scientifically invalid without cross-method validation. Just as statistical significance requires specifying the test statistic, stability must be validated across multiple attribution paradigms or explicitly scoped to a single method’s computational objective. In controlled chest X-ray experiments, DenseNet201, ResNet50V2, and InceptionV3 achieve $>$99% AUC but exhibit reversed stability rankings across attribution methods. LayerCAM ranks InceptionV3 highest (IoU 0.777), while GradCAM++ favors DenseNet201, reducing InceptionV3’s score by 17.3%. These findings establish that explanation stability is an emergent property of the model–method pair, not an intrinsic model trait. We argue that explanation-based claims should be validated across multiple attribution methods and urge that regulatory submissions explicitly specify attribution operators to avoid illusory safety assurances.

More from ICML 2026

All ICML 2026 papers →