ASSS: A Differentiable Adversarial Framework for Task-Aware Data Reduction

TL;DR

ASSS is a differentiable adversarial framework that optimizes data reduction by selecting task-relevant samples, achieving high performance with significantly less data.

Contribution

It introduces a novel end-to-end differentiable method for task-aware data reduction based on a minimax game and Gumbel-Softmax relaxation.

Findings

Achieves 98.9% performance retention with only 30% of data

Outperforms random, K-means, and gradient-based sampling methods

Retains samples near decision boundaries as visualized

Abstract

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. Using Gumbel-Softmax relaxation, ASSS enables end-to-end gradient flow and is theoretically grounded in the information bottleneck principle. Experiments on multiple benchmarks show that ASSS achieves a performance retention rate (PRR) of 98.9% while using only 30% of the data, significantly outperforming random sampling, K-means, and gradient-based methods. Visualizations confirm that ASSS preferentially retains samples near decision boundaries. The framework is scalable, fully differentiable, and easily integrated into existing training pipelines. This work introduces a new paradigm for task-aware data reduction that directly optimizes subset selection for the downstream objective, offering a principled and practical solution to the scalability challenges in modern deep learning.