arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2610.13129stat.MLcs.LG
收藏

基于梯度的连续动态时间规整快速近似方法

Fast Gradient-Based Approximation of Continuous Dynamic Time Warping

INRIA Rennes(法国国家信息与自动化研究所雷恩中心);IRISA, UMR 6074, CNRS, Université de Rennes(雷恩大学);IRISA, UMR 6074, CNRS, Université de Bretagne Sud(南布列塔尼大学)

Pierre Houedry, Romain Tavenard, Nicolas Courty, Romaric Gaudel

AI总结:

本文提出FastCDTW,一种高效的连续动态时间规整近似求解器,含两种计算策略,可适配GPU,集成于时间序列平均、分类等任务验证了优势。

摘要:

动态时间规整(DTW)是衡量时间序列差异的常用方法,它通过动态规划寻找时间戳间的最优规整路径,但速度较慢且不可微。为此已提出多种变体,其中Soft-DTW给出了可计算梯度的平滑DTW形式,连续动态时间规整(CDTW)则通过可参数化的规整函数引入了连续形式。不过这两种方法的计算成本仍很高,在现代机器学习流程中无法适用于大型数据集。本文从多个角度重新研究CDTW:在确立其在温和条件下可微后,提出FastCDTW,这是一种高效的近似求解器,包含两种不同计算策略:针对分段线性规整函数的精确闭式方案,以及针对平滑规整函数(如隐式神经网络)的无偏蒙特卡洛估计器。两种策略均能提供关于规整函数和输入时间序列的梯度,且生成的算法适配GPU。我们通过将其集成到更复杂的学习任务中,包括时间序列平均、分类和深度领域自适应,验证了其优势。

英文摘要:

Dynamic Time Warping (DTW) is a common choice for measuring the discrepancy between time series. It proceeds by finding an optimal warping between timestamps through dynamic programming, which can be slow and is not differentiable. Consequently, several variants have been proposed. Among them, Soft-DTW proposes a smooth formulation of DTW, amenable to gradient computation, while Continuous-DTW (CDTW) introduces a continuous formulation through parametrizable warping functions. Yet, computational costs remain high for both methods and impractical for large datasets in modern machine learning pipelines. This paper revisits CDTW in several ways. After establishing its differentiability under mild conditions, we introduce FastCDTW, which is an efficient approximate solver with two different computational strategies: an exact closed-form scheme for piecewise-linear warping functions, and an unbiased Monte Carlo estimator for smooth ones, such as implicit neural networks. Both strategies provide gradients with respect to the warping function and the input time series, and the resulting algorithm is GPU-friendly. We demonstrate its benefits through its integration into more complex learning problems, including time series averaging, classification and deep domain adaptation.

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。

↑