通过建模多视觉模式分布实现平衡的数据集蒸馏
Balanced Dataset Distillation via Modeling Multiple Visual Pattern Distribution
把「每类单簇」的假设换成类内多视觉模式分布，用层次语义结构在中心与边缘各取子集做平衡 coreset，可直接迁移到视觉数据集压缩与高效训练流程中，并提供了信息覆盖与性能一致性的理论保证。
原文依据 BPS models the multiple visual pattern distribution within each class via a hierarchical semantic structure inherent to the dataset.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论数据集蒸馏（DD）旨在将大规模数据集压缩为每类少量（IPC）的浓缩图像，从而实现高效的网络训练。此前的 coreset 选择方法与基于合成的 DD 方法取得了合理性能。然而，我们的深入研究发现现有方法存在一个共同问题：模式不平衡。具体而言，它们要么过度强调代表每类中大多数的类通用模式，要么聚焦于对模型泛化至关重要的少数边缘模式。为解决该问题，我们提出了一种新框架——平衡模式选择（BPS）。与假设每个类构成单一簇的先验方法不同，BPS 借助数据集固有的层次语义结构来建模每个类内的多视觉模式分布，随后在每个模式的中心（类通用模式）与边缘（边缘模式）以平衡方式选取两个互补子集，从而产生模式平衡的 coreset。在理论上，我们证明 BPS 所选 coreset 在信息覆盖与性能两方面均与原始数据集一致。此外，其与模型无关的选择特性确保了跨架构泛化能力，而「一次优化适配所有 IPC」的性质保证了效率。在四个基准上的大量实验表明，BPS 显著优于现有最优方法。源代码见：https://github.com/BeCarefulOfYournaoke/BPS。
英文原文摘要 1504 字符
Dataset Distillation (DD) aims to compress large-scale datasets into a small number of condensed Images Per Class (IPC), enabling efficient network training. Previous coreset selection and synthetic-based DD methods achieve reasonable performance. However, our in-depth investigation reveals that existing methods share a common issue: pattern imbalance. Specifically, they either overemphasize class-general patterns representing the majority of each class or focus on fewer marginal patterns critical for model generalization. To address this issue, we propose a novel framework, Balanced Patterns Selection (BPS). Unlike prior methods that assume each class forms a single cluster, BPS models the multiple visual pattern distribution within each class via a hierarchical semantic structure inherent to the dataset. It then selects two complementary subsets in a balanced manner from the center (class-general patterns) and the margins (marginal patterns) of each pattern, producing a pattern-balanced coreset. Theoretically, we prove that the BPS-selected coreset aligns with the original dataset in both information coverage and performance. Moreover, its model-agnostic selection nature ensures cross-architecture generalization, while the Optimize-Once-for-All-IPCs property guarantees efficiency. Extensive experiments on four benchmarks demonstrate that BPS significantly outperforms existing state-of-the-art methods. The source code is available at: https://github.com/BeCarefulOfYournaoke/BPS.