HyperCore: Coreset Selection under Noise via Hypersphere Models

2025-09-25

AI Summary

Existing core-set selection methods neglect label noise and rely on predefined pruning ratios, limiting their practicality. This paper proposes a noise-aware hyperspherical core-set selection framework: for each class, it constructs a lightweight hyperspherical model to embed samples and identify anomalies based on intra-class centroid distances. It further introduces Youden’s J statistic to adaptively determine the pruning threshold—eliminating the need for manual hyperparameter tuning while automatically filtering mislabeled and ambiguous samples, even under low-data and high-noise conditions. Experiments demonstrate that our method significantly outperforms state-of-the-art approaches across diverse noise settings, yielding more compact and information-rich subsets. These subsets effectively support robust and scalable downstream learning tasks.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

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.

Problem

• Selects representative data subsets despite annotation errors
• Adaptively determines pruning thresholds without manual tuning
• Identifies and removes mislabeled samples for robust learning

Innovation

• Uses hypersphere models for class representation
• Adaptively selects thresholds via Youden's statistic
• Automatically prunes noisy data without hyperparameter tuning