HyperCore: Coreset Selection under Noise via Hypersphere Models

Sep 26, 2025·
Brian Bernhard Moser
,
Arundhati Shanbhag
Tobias Christian Nauen Tobias Christian Nauen
,
Stanislav Frolov
,
Federico Raue
,
Joachim Folz
,
Andreas Dengel
PDF
Abstract
The goal of coreset selection methods is to identify representative subsets of datasets for efficient model training. Yet, existing methods often ignore the possibility of annotation errors and require fixed pruning ratios, making them impractical in real-world settings. We present HyperCore, a robust and adaptive coreset selection framework designed explicitly for noisy environments. HyperCore leverages lightweight hypersphere models learned per class, embedding in-class samples close to a hypersphere center while naturally segregating out-of-class samples based on their distance. By using Youden’s J statistic, HyperCore can adaptively select pruning thresholds, enabling automatic, noise-aware data pruning without hyperparameter tuning. Our experiments reveal that HyperCore consistently surpasses state-of-the-art coreset selection methods, especially under noisy and low-data regimes. HyperCore effectively discards mislabeled and ambiguous points, yielding compact yet highly informative subsets suitable for scalable and noise-free learning.
Type
Publication
arXiv preprint (arXiv)
publications

For more information, see the paper pdf.

Citation

If you use this work, please cite our paper:

@misc{moser2025hypercore,
  title = {HyperCore: Coreset Selection under Noise via Hypersphere Models},
  author = {Brian B. Moser and Arundhati S. Shanbhag and Tobias C. Nauen and
            Stanislav Frolov and Federico Raue and Joachim Folz and Andreas
            Dengel},
  year = {2025},
  eprint = {2509.21746},
  archivePrefix = {arXiv},
  primaryClass = {cs.LG},
  url = {https://arxiv.org/abs/2509.21746},
}
Deep Learning Coreset Selection
Tobias Christian Nauen
Authors
Tobias Christian Nauen
PhD Student
I’m a researcher of artificial intelligence at DFKI and RPTU Kaiserslautern-Landau. My research interests include efficient deep learning, transformer models, multimodal learning, and computer vision. In my PhD project, my focus lies on the development of efficient transformer models for vision, language, and multimodal tasks.