arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2409.16239cs.CVcs.AI
收藏

标签增强的数据集蒸馏

Label-Augmented Dataset Distillation

Yonsei University(延世大学);KAIST AI(韩国科学技术院人工智能研究院)

Seoungyoon Kang, Youngsun Lim, Hyunjung Shim

AI总结:

提出标签增强的数据集蒸馏框架LADD,通过对合成图像子采样生成密集标签,在极小存储开销下显著提升训练效率、准确率及跨架构鲁棒性。

摘要:

传统的数据集蒸馏主要关注图像表示,而往往忽略了标签的重要作用。在本研究中,我们提出了标签增强的数据集蒸馏(LADD),这是一种通过标签增强来提升数据集蒸馏效果的新框架。LADD对每张合成图像进行子采样,生成额外的密集标签以捕捉丰富的语义信息。这些密集标签仅需增加2.5%的存储空间(ImageNet子集),却能带来显著的性能提升,提供强有力的学习信号。我们的标签生成策略可与现有的数据集蒸馏方法互补,显著提升其训练效率和性能。实验结果表明,LADD在计算开销和准确率方面均优于现有方法。结合三种高性能数据集蒸馏算法,LADD实现了平均14.9%的准确率提升。此外,我们方法的有效性在多种数据集、蒸馏超参数和算法上均得到验证。最后,我们的方法提升了蒸馏数据集的跨架构鲁棒性,这在应用场景中具有重要意义。

英文摘要:

Traditional dataset distillation primarily focuses on image representation while often overlooking the important role of labels. In this study, we introduce Label-Augmented Dataset Distillation (LADD), a new dataset distillation framework enhancing dataset distillation with label augmentations. LADD sub-samples each synthetic image, generating additional dense labels to capture rich semantics. These dense labels require only a 2.5% increase in storage (ImageNet subsets) with significant performance benefits, providing strong learning signals. Our label generation strategy can complement existing dataset distillation methods for significantly enhancing their training efficiency and performance. Experimental results demonstrate that LADD outperforms existing methods in terms of computational overhead and accuracy. With three high-performance dataset distillation algorithms, LADD achieves remarkable gains by an average of 14.9% in accuracy. Furthermore, the effectiveness of our method is proven across various datasets, distillation hyperparameters, and algorithms. Finally, our method improves the cross-architecture robustness of the distilled dataset, which is important in the application scenario.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑