Group-invariant Coresets for Data-efficient Active Learning
2026-07-01 • Machine Learning
Machine Learning
AI summaryⓘ
The authors present GRINCO, a method that helps machines pick the most useful examples to learn from by recognizing when data points are just transformed versions of the same thing (like rotated images). Instead of treating these similar versions separately, GRINCO groups them together and selects examples based on these groups, saving effort. They also provide a theoretical explanation of why this approach works well. In tests, their method required fewer labeled examples to learn effectively compared to traditional methods, especially when there was a lot of similar, transformed data.
active learningcoresetgroup invariancequotient spacetransformation groupk-center selectioninvariant embeddinglabel efficiencygeneralization boundorbit coverage
Authors
L. C. Ayres, J. C. M. Bermudez, S. J. M. de Almeida, R. A. Borsoi
Abstract
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.