English

Adaptive Dataset Quantization: A New Direction for Dataset Pruning

Computer Vision and Pattern Recognition 2025-12-09 v1 Artificial Intelligence

Abstract

This paper addresses the challenges of storage and communication costs for large-scale datasets in resource-constrained edge devices by proposing a novel dataset quantization approach to reduce intra-sample redundancy. Unlike traditional dataset pruning and distillation methods that focus on inter-sample redundancy, the proposed method compresses each image by reducing redundant or less informative content within samples while preserving essential features. It first applies linear symmetric quantization to obtain an initial quantization range and scale for each sample. Then, an adaptive quantization allocation algorithm is introduced to distribute different quantization ratios for samples with varying precision requirements, maintaining a constant total compression ratio. The main contributions include: (1) being the first to use limited bits to represent datasets for storage reduction; (2) introducing a dataset-level quantization algorithm with adaptive ratio allocation; and (3) validating the method's effectiveness through extensive experiments on CIFAR-10, CIFAR-100, and ImageNet-1K. Results show that the method maintains model training performance while achieving significant dataset compression, outperforming traditional quantization and dataset pruning baselines under the same compression ratios.

Keywords

Cite

@article{arxiv.2512.05987,
  title  = {Adaptive Dataset Quantization: A New Direction for Dataset Pruning},
  author = {Chenyue Yu and Jianyu Yu},
  journal= {arXiv preprint arXiv:2512.05987},
  year   = {2025}
}

Comments

Accepted by ICCPR 2025

R2 v1 2026-07-01T08:12:12.073Z