arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2003.06462stat.MEcs.CG
收藏

时间序列分类的持久同调方法

A Persistent Homology Approach to Time Series Classification

Yu-Min Chung, William Cruse, Austin Lawson

摘要:

拓扑数据分析(TDA)是计算拓扑学的一个新兴领域,它可以通过持久同源性来观察数据集的拓扑结构。通过考虑一系列的子水平集,得到一个过滤,跟踪拓扑信息的变化。这些更改可以记录在称为{ \it持久性图}的多集合中。将存储在持久性图中的信息转换为与现代机器学习算法兼容的形式是TDA研究的主要方向。{ \it Persistence curves}是最近开发的一个框架,它提供了一种规范而灵活的方法来将持久性图中呈现的信息编码为向量。在这项工作中,我们提出了一套新的指标持久性曲线的基础上。我们证明了所提出的度量的稳定性。最后,我们将这些指标应用于UCR时间序列分类存档。这些实证结果表明,在大多数情况下,我们的指标表现优于相关的基准,值得进一步研究。

英文摘要:

Topological Data Analysis (TDA) is a rising field of computational topology in which the topological structure of a data set can be observed by persistent homology. By considering a sequence of sublevel sets, one obtains a filtration that tracks changes in topological information. These changes can be recorded in multi-sets known as {\it persistence diagrams}. Converting information stored in persistence diagrams into a form compatible with modern machine learning algorithms is a major vein of research in TDA. {\it Persistence curves}, a recently developed framework, provides a canonical and flexible way to encode the information presented in persistence diagrams into vectors. In this work, we propose a new set of metrics based on persistence curves. We prove the stability of the proposed metrics. Finally, we apply these metrics to the UCR Time Series Classification Archive. These empirical results show that our metrics perform better than the relevant benchmark in most cases and warrant further study.

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。

↑