Pattern Recognition ( IF 7.5 ) Pub Date : 2021-02-16 , DOI: 10.1016/j.patcog.2021.107895 Geoffrey I. Webb , François Petitjean
Dynamic Time Warping () is a popular similarity measure for aligning and comparing time series. Due to ’s high computation time, lower bounds are often employed to screen poor matches. Many alternative lower bounds have been proposed, providing a range of different trade-offs between tightness and computational efficiency. provides a useful trade-off in many applications. Two recent lower bounds, and are substantially tighter than . All three have the same worst case computational complexity—linear with respect to series length and constant with respect to window size. We present four new lower bounds in the same complexity class. is substantially tighter than with only modest additional computational overhead. is more efficient than while often providing a tighter bound. is always tighter than . The parameter free is usually tighter than . A parameterized variant, LB_Webb_Enhanced, is always tighter than . A further variant, is useful for some constrained distance functions. In extensive experiments, proves to be very effective for nearest neighbor search.
中文翻译:
严格的下限,可动态调整时间
动态时间规整()是用于对齐和比较时间序列的一种流行的相似性度量。由于的运算时间长,下限通常用于筛选不匹配的内容。已经提出了许多替代的下限,从而在紧密度和计算效率之间提供了一系列不同的权衡。在许多应用程序中提供了有用的折衷。最近的两个下限, 和 比 。这三个变量都具有相同的最坏情况的计算复杂性-相对于序列长度是线性的,而相对于窗口大小是恒定的。我们提出了四个新的 同一复杂度类别中的下界。 比 仅需少量的额外计算开销。 比 同时通常会提供更严格的界限。 总是比 。无参数 通常比 。参数化变体LB_Webb_Enhanced总是比。另一个变体对于某些约束距离函数很有用。在广泛的实验中 事实证明,它对于最近邻居搜索非常有效。