arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2408.13189cs.LGcs.AI
收藏

利用几何信息加速 k-means++ 算法

Accelerating the k-means++ Algorithm by Using Geometric Information

Universitat Politècnica de Catalunya (UPC)(加泰罗尼亚理工大学);Artificial Intelligence Research Institute (IIIA-CSIC)(人工智能研究所)

Guillem Rodríguez Corominas, Maria J. Blesa, Christian Blum

AI总结:

本文提出利用三角不等式、范数过滤器和两步采样加速精确 k-means++ 算法,实验证明其在减少距离计算上优于标准版本,且在高维和低维数据及多任务下均表现良好。

摘要:

本文提出了一种利用几何信息的精确 k-means++ 算法加速方法,具体使用了三角不等式和额外的范数过滤器,并结合了两步采样过程。实验表明,该加速版本在访问点数和距离计算方面优于标准 k-means++ 版本,且随着簇数量的增加,加速效果更显著。利用三角不等式的版本对低维数据尤为有效,而基于范数的额外过滤器则增强了在点间范数方差较大的高维实例中的性能。额外实验展示了算法在多个作业并发执行时的行为,并考察了内存性能对实际加速的影响。

英文摘要:

In this paper, we propose an acceleration of the exact k-means++ algorithm using geometric information, specifically the Triangle Inequality and additional norm filters, along with a two-step sampling procedure. Our experiments demonstrate that the accelerated version outperforms the standard k-means++ version in terms of the number of visited points and distance calculations, achieving greater speedup as the number of clusters increases. The version utilizing the Triangle Inequality is particularly effective for low-dimensional data, while the additional norm-based filter enhances performance in high-dimensional instances with greater norm variance among points. Additional experiments show the behavior of our algorithms when executed concurrently across multiple jobs and examine how memory performance impacts practical speedup.

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。

↑