HyperCore: Coreset Selection under Noise via Hypersphere Models
This addresses the challenge of efficient and robust model training for machine learning practitioners by providing an adaptive method that handles annotation errors, though it is incremental as it builds on existing coreset selection approaches.
The paper tackles the problem of coreset selection in noisy datasets by introducing HyperCore, a framework that uses hypersphere models and adaptive thresholds to prune mislabeled data, resulting in improved performance over state-of-the-art methods, especially in noisy and low-data settings.
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.