Surprise-Annealed Dual-Tail Loss for Robust Noisy Labels Learning
Entropy · 19.11.2026 · 28 (11) · 1279
DOI: 10.3390/e28111279 ↗Kurz gefasst
Fachartikel in der Zeitschrift Entropy, online erschienen am 19.11.2026. Verfasst von Qian Zhang und 3 weiteren Autor:innen. Die Arbeit nennt 40 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: the Qing Lan Project of Jiangsu Province, the National Natural Science Foundation of China.
Abstract (Originalsprache)
Overparameterized deep neural networks (DNNs) rely heavily on high-quality labels, while conventional cross entropy (CE) tends to overfit incorrect annotations under label noise, leading to degraded generalization performance. Robust loss functions alleviate this issue by reducing sensitivity to noisy labels. However, strict symmetric losses often exhibit limited fitting capability, while full normalization may alter the optimization properties of the original loss. To address these limitations, we propose surprise-annealed dual-tail loss (SADT), a bounded multiclass classification objective constructed from predictive surprise. Specifically, SADT retains the original predictive surprise in the active score construction when the predicted probability is at least the chance level, progressively compresses moderate surprise through an inverse hyperbolic sine branch, and bounds the extreme tail using an arctangent branch. By combining a normalized active score with a log balanced passive score, SADT achieves an exact constant class sum. Theoretically, we establish its boundedness and prove that it preserves clean risk minimizers under uniform symmetric label noise with a corruption rate below (C − 1)/C. For class conditional and instance-dependent noise, we derive conditional robustness guarantees and an excess risk bound. Experiments show that SADT improves accuracy over NCE + NNCE by 2.64% and 3.16% on CIFAR-10 with 80% symmetric noise and WebVision, respectively, while showing weaker performance on Clothing1M. These results demonstrate the effectiveness of SADT under several challenging noise conditions while revealing its limitations in learning from difficult clean samples.
Förderung
- the Qing Lan Project of Jiangsu Province
- the National Natural Science Foundation of China
Zitieren
APA
Zhang, Q., Chen, Y., Yuan, Y., & Jin, G. (2026). Surprise-Annealed Dual-Tail Loss for Robust Noisy Labels Learning. Entropy, 28(11), 1279. https://doi.org/10.3390/e28111279
Vancouver
Zhang Q, Chen Y, Yuan Y, Jin G. Surprise-Annealed Dual-Tail Loss for Robust Noisy Labels Learning. Entropy. 2026;28(11):1279. doi:10.3390/e28111279
BibTeX
@article{Zhang2026,
author = {Zhang, Qian and Chen, Yanzhan and Yuan, Yongqi and Jin, Ge},
title = {Surprise-Annealed Dual-Tail Loss for Robust Noisy Labels Learning},
journal = {Entropy},
year = {2026},
volume = {28},
number = {11},
pages = {1279},
doi = {10.3390/e28111279}
}
EUVIMED