SAMix: Calibrated and Accurate Continual Learning via Sphere-Adaptive Mixup and Neural Collapse
Machine Learning · 2026-07 · 115 (7)
DOI: 10.1007/s10994-026-07049-3 ↗Kurz gefasst
Fachartikel in der Zeitschrift Machine Learning, online erschienen am 03.07.2026. Verfasst von Trung-Anh Dang und 3 weiteren Autor:innen. Die Arbeit nennt 78 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Université d' Orleans.
Abstract (Originalsprache)
While most continual learning methods focus on mitigating forgetting and improving accuracy, they often overlook the critical aspect of network calibration, despite its importance. Neural collapse, a phenomenon where last-layer features collapse to their class means, has demonstrated advantages in continual learning by reducing feature-classifier misalignment. Few works aim to improve the calibration of continual models for more reliable predictions. Our work goes a step further by proposing a novel method that not only enhances calibration but also improves performance by reducing overconfidence, mitigating forgetting, and increasing accuracy. We introduce Sphere-Adaptive Mixup (SAMix), an adaptive mixup strategy tailored for neural collapse-based methods. SAMix adapts the mixing process to the geometric properties of feature spaces under neural collapse, ensuring more robust regularization and alignment. Experiments show that SAMix significantly boosts performance, surpassing SOTA methods in continual learning while also improving model calibration. SAMix enhances both across-task accuracy and the broader reliability of predictions, making it a promising advancement for robust continual learning systems.
Förderung
- Université d' Orleans
Zitieren
APA
Dang, T., Nguyen, V., Vu, N., & Vrain, C. (2026). SAMix: Calibrated and Accurate Continual Learning via Sphere-Adaptive Mixup and Neural Collapse. Machine Learning, 115(7). https://doi.org/10.1007/s10994-026-07049-3
Vancouver
Dang T, Nguyen V, Vu N, Vrain C. SAMix: Calibrated and Accurate Continual Learning via Sphere-Adaptive Mixup and Neural Collapse. Machine Learning. 2026;115(7). doi:10.1007/s10994-026-07049-3
BibTeX
@article{Dang2026,
author = {Dang, Trung-Anh and Nguyen, Vincent and Vu, Ngoc-Son and Vrain, Christel},
title = {SAMix: Calibrated and Accurate Continual Learning via Sphere-Adaptive Mixup and Neural Collapse},
journal = {Machine Learning},
year = {2026},
volume = {115},
number = {7},
doi = {10.1007/s10994-026-07049-3}
}
EUVIMED