提出分阶段筛选策略，让合成数据与真实数据更好配合，大幅提高高压缩比下的图像分类效果。
Curriculum Coarse-to-Fine Selection for High-IPC Dataset Distillation
- 采用课程式分阶段筛选真实数据，逐步匹配当前合成数据的分布。
- 在高每类图像数（high-IPC）设置下，比现有方法提升最多6.6%准确率。
- 适合需要高压缩比且保持高精度的模型训练场景，如资源受限部署。
数据蒸馏（DD）在每类图像数较少（IPC）时表现优异，但在高IPC场景下效果下降。已有研究发现，将合成数据与真实数据结合可缓解性能衰减，但现有一次性独立选择机制导致合成与真实数据不兼容。为此，本文提出一种新的课程式粗到精筛选（CCFS）方法，用于高效高IPC数据蒸馏。CCFS采用课程化框架筛选真实数据，通过粗到精策略根据当前合成数据分布逐阶段选择合适的真实样本。大量实验验证了其有效性，在高IPC设置下，于CIFAR-10上超越当前最优方法6.6%，CIFAR-100上提升5.8%，Tiny-ImageNet上提升3.4%。特别地，使用ResNet-18在仅保留Tiny-ImageNet 20%数据量的情况下，达到60.2%测试准确率，仅比全量训练低0.3%。代码已开源。
原文摘要 · Abstract (English)
Dataset distillation (DD) excels in synthesizing a small number of images per class (IPC) but struggles to maintain its effectiveness in high-IPC settings. Recent works on dataset distillation demonstrate that combining distilled and real data can mitigate the effectiveness decay. However, our analysis of the combination paradigm reveals that the current one-shot and independent selection mechanism induces an incompatibility issue between distilled and real images. To address this issue, we introduce a novel curriculum coarse-to-fine selection (CCFS) method for efficient high-IPC dataset distillation. CCFS employs a curriculum selection framework for real data selection, where we leverage a coarse-to-fine strategy to select appropriate real data based on the current synthetic dataset in each curriculum. Extensive experiments validate CCFS, surpassing the state-of-the-art by +6.6\% on CIFAR-10, +5.8\% on CIFAR-100, and +3.4\% on Tiny-ImageNet under high-IPC settings. Notably, CCFS achieves 60.2\% test accuracy on ResNet-18 with a 20\% compression ratio of Tiny-ImageNet, closely matching full-dataset training with only 0.3\% degradation. Code: https://github.com/CYDaaa30/CCFS.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。