Researchers explore data symmetries to improve noisy dataset selection for ML

Researchers have developed a new method to identify optimal subsets of training data, particularly when dealing with label noise. This approach leverages data symmetries and invariance properties to improve the accuracy of k-nearest neighbors (k-NN) in selecting low-noise samples. The findings suggest that exploiting these underlying symmetries can lead to performance comparable to training on noise-free datasets, even in high-dimensional settings. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Improves robustness of models trained on potentially noisy real-world datasets.

RANK_REASON Academic paper detailing a novel method for data selection in machine learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• k-nearest neighbors
• Bayes optimal classifier

• paper
• other