arXiv 2602.090082026-02-09cs.LG

ShapeCond: 快速形状图引导的数据集压缩用于时间序列分类

ShapeCond: Fast Shapelet-Guided Dataset Condensation for Time Series Classification

Shanghai Key Lab of Data Science, College of Computer Science and Artificial Intelligence, Fudan University, Shanghai, China（上海数据科学联合实验室，计算机科学与人工智能学院，复旦大学，上海，中国）；Stanford University, Stanford, USA（斯坦福大学，斯坦福，美国）；Ocean University of China（中国海洋大学）；Shanghai Data Research Institute（上海数据研究院）；Mohamed bin Zayed University of Artificial Intelligence Abu Dhabi, United Arab Emirates（阿布扎克穆罕默德·本·扎耶德人工智能大学）

Sijia Peng, Yun Xiong, Xi Chen, Yi Xie, Guanzhi Li, Yanwei Yu, Yangyong Zhu, Zhiqiang Shen

备注：: Code at: https://github.com/lunaaa95/ShapeCond

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

ShapeCond通过形状图引导的优化策略，高效压缩时间序列数据集，提升下游分类准确性。

摘要：

时间序列数据支持许多领域（例如金融和气候科学），但其快速增长会加剧存储和计算的压力。数据集压缩可以通过合成一个紧凑的训练集来缓解这一问题，该训练集保留了关键信息。然而，大多数压缩方法都是图像导向的，通常在时间序列上失败，因为它们忽略了时间序列特有的时间结构，特别是局部判别模式如形状图。在本文中，我们提出ShapeCond，一种新的、高效的用于时间序列分类的数据集压缩框架，通过基于形状图的数据集知识，利用形状图引导的优化策略。我们的形状图辅助合成成本与序列长度无关：更长的序列会产生更大的合成速度提升（例如，比先前最先进的方法CondTSC在时间序列压缩上的速度提高29倍，以及在Sleep数据集上使用形状图时，达到10,000倍的提升）。通过显式保留关键的局部模式，ShapeCond提高了下游的准确性，并在广泛的实验中始终优于所有先前最先进的时间序列数据集压缩方法。代码可在https://github.com/lunaaa95/ShapeCond上获得。

英文摘要：

Time series data supports many domains (e.g., finance and climate science), but its rapid growth strains storage and computation. Dataset condensation can alleviate this by synthesizing a compact training set that preserves key information. Yet most condensation methods are image-centric and often fail on time series because they miss time-series-specific temporal structure, especially local discriminative motifs such as shapelets. In this work, we propose ShapeCond, a novel and efficient condensation framework for time series classification that leverages shapelet-based dataset knowledge via a shapelet-guided optimization strategy. Our shapelet-assisted synthesis cost is independent of sequence length: longer series yield larger speedups in synthesis (e.g., 29$\times$ faster over prior state-of-the-art method CondTSC for time-series condensation, and up to 10,000$\times$ over naively using shapelets on the Sleep dataset with 3,000 timesteps). By explicitly preserving critical local patterns, ShapeCond improves downstream accuracy and consistently outperforms all prior state-of-the-art time series dataset condensation methods across extensive experiments. Code is available at https://github.com/lunaaa95/ShapeCond.

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。