基于最近邻搜索的 Earth Mover's Distance 高效近似
Efficient approximation of Earth Mover's Distance Based on Nearest Neighbor Search
AI总结:
本文提出 NNS-EMD,利用最近邻搜索与 GPU 向量化高效近似 EMD,在图像分类、检索和颜色迁移中实现高精度、低内存占用及最高 135 倍加速。
摘要:
Earth Mover's Distance(EMD,推土机距离)是两个分布之间的一种重要相似性度量,应用于计算机视觉和许多其他应用领域。然而,其精确计算在计算和内存方面开销很大,这限制了它在大规模问题中的可扩展性和适用性。人们已经提出了多种近似 EMD 算法以降低计算成本,但这些算法会带来较低的精度,并且可能需要额外的内存使用或人工参数调优。本文提出一种名为 NNS-EMD 的新方法,利用 Nearest Neighbor Search(NNS,最近邻搜索)来近似 EMD,以实现高精度、低时间复杂度和高内存效率。NNS 操作减少了每次 NNS 迭代中需要比较的数据点数量,并为并行处理提供了机会。我们进一步通过 GPU 上的向量化来加速 NNS-EMD,这对大型数据集尤其有利。我们在图像分类和检索任务上将 NNS-EMD 与精确 EMD 以及最先进的近似 EMD 算法进行比较。我们还应用 NNS-EMD 来计算传输映射,并实现图像之间的颜色迁移。NNS-EMD 的速度可比精确 EMD 实现快 44 倍至 135 倍,并且相较于现有近似 EMD 方法,在精度、加速比和内存效率方面均表现更优。
英文摘要:
Earth Mover's Distance (EMD) is an important similarity measure between two distributions, used in computer vision and many other application domains. However, its exact calculation is computationally and memory intensive, which hinders its scalability and applicability for large-scale problems. Various approximate EMD algorithms have been proposed to reduce computational costs, but they suffer lower accuracy and may require additional memory usage or manual parameter tuning. In this paper, we present a novel approach, NNS-EMD, to approximate EMD using Nearest Neighbor Search (NNS), in order to achieve high accuracy, low time complexity, and high memory efficiency. The NNS operation reduces the number of data points compared in each NNS iteration and offers opportunities for parallel processing. We further accelerate NNS-EMD via vectorization on GPU, which is especially beneficial for large datasets. We compare NNS-EMD with both the exact EMD and state-of-the-art approximate EMD algorithms on image classification and retrieval tasks. We also apply NNS-EMD to calculate transport mapping and realize color transfer between images. NNS-EMD can be 44x to 135x faster than the exact EMD implementation, and achieves superior accuracy, speedup, and memory efficiency over existing approximate EMD methods.
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。