AI RESEARCH
ASSS: A Differentiable Adversarial Framework for Task-Aware Data Reduction
arXiv CS.LG
•
ArXi:2601.02081v3 Announce Type: replace Massive datasets often contain redundancy that inflates computational costs without improving generalization. Existing data reduction methods are typically task-agnostic, discarding informative boundary samples and yielding suboptimal performance. We propose Adversarial Soft-Selection Subsampling (ASSS), a differentiable framework that casts data reduction as a minimax game between a learnable selector and a task network.