Subtype Robustness Is Not Just Accuracy: Calibration Under Unseen Subtype Shift

Excerpt

arXiv:2608.00928v2 Announce Type: replace Abstract: Subtype robustness asks whether a model keeps the correct coarse prediction when test examples come from fine-grained subtypes absent from training but still inside a known coarse category. Prior work studies this almost entirely through accuracy. We ask whether the model also stays calibrated. We present the first systematic study of the question across ImageNet, BREEDS, iNaturalist and CIFAR-100 with five architectures. Calibration breaks dow