Dataset Pruning from First Principles: A Label-Free Linear Programming Approach

Dataset pruning reduces a large training set to a representative subset while preserving model performance. Existing geometry-based methods typically assume that nearby points in embedding space share similar properties.