Search papers, labs, and topics across Lattice.
This paper introduces GRINCO, a group-invariant coreset framework designed to enhance data-efficient active learning by leveraging known data symmetries to minimize labeling costs. By querying samples in a quotient space defined by transformation groups, GRINCO selects data based on orbits rather than individual instances, leading to improved label efficiency and better coverage of the data distribution. Experimental results demonstrate that GRINCO outperforms traditional coreset methods, particularly in scenarios with significant group-induced redundancy, achieving stronger performance on both synthetic and image datasets.
GRINCO achieves superior label efficiency by intelligently querying data in a transformed space, effectively addressing the redundancy problem in active learning.
Active learning reduces labeling cost by querying the most informative unlabeled samples, but standard coreset methods ignore known data symmetries and can waste budget on transformed versions of the same instance. We propose GRINCO, a group-invariant coreset framework that performs acquisition in the quotient space induced by a transformation group, so that selection operates on orbits rather than raw samples. The method uses either canonical representatives or learned orbit-separating invariant embeddings to define practical quotient metrics, and combines quotient-space k-center selection with invariant training through an orbit-averaged loss. We further derive a generalization bound that relates excess orbit-averaged risk to quotient-space coverage, label uncertainty, and intra-orbit variability. Experiments on synthetic scale-invariant data and image benchmarks with rotation-induced redundancy show that GRINCO improves orbit coverage and achieves stronger label efficiency than conventional coreset baselines, especially when group-induced redundancy is substantial.
