NVIDIA cuML 多 GPU UMAP：大规模降维提速至分钟级
原标题：Run Massive-Scale UMAP in Minutes Using Multiple GPUs—Without Losing Accuracy
AI 摘要
NVIDIA cuML 和 cuVS 25.06 版本引入了多 GPU 分布式 UMAP 功能，通过将昂贵的全邻居 kNN 图构建步骤分布到多个 GPU 上，显著提升了大规模数据集的降维性能。该方法基于先前提出的 out-of-core 技术，将数据集划分为平衡簇并重叠向量，每个 GPU 独立计算局部 kNN 图并合并，避免了昂贵的全对全通信。实验表明，该功能可在几分钟内处理数百 GB 的数据，而无需牺牲嵌入质量。
正文节选
Uniform Manifold Approximation and Projection (UMAP) is a dimensionality reduction technique widely used for visualization and feature extraction. Applications range across exploratory data analysis, topic modeling, and single-cell analysis. Many of these workflows are iterative and exploratory, requiring UMAP to be run repeatedly as users analyze their data or tune parameters. As datasets grow, the cost of each UMAP run increases substantially, making interactive exploration and iterative analy