Differentiable Divergences Between Time Series

Abstract

Computing the discrepancy between time series of variable sizes is notoriously challenging. While dynamic time warping (DTW) is popularly used for this purpose, it is not differentiable everywhere and is known to lead to bad local optima when used as a "loss". Soft-DTW addresses these issues, but it is not a positive definite divergence: due to the bias introduced by entropic regularization, it can be negative and it is not minimized when the time series are equal. We propose in this paper a new divergence, dubbed soft-DTW divergence, which aims to correct these issues. We study its properties; in particular, under conditions on the ground cost, we show that it is a valid divergence: it is non-negative and minimized if and only if the two time series are equal. We also propose a new "sharp" variant by further removing entropic bias. We showcase our divergences on time series averaging and demonstrate significant accuracy improvements compared to both DTW and soft-DTW on 84 time series classification datasets.

arXiv commentV3: AISTATS 2021 camera-ready

References in corpus (8)

• A kernel for time series based on global alignments
• Why Delannoy numbers?
• Interpolating between Optimal Transport and MMD using Sinkhorn Divergences
• Differentiable Dynamic Programming for Structured Prediction and Attention
• Differential Properties of Sinkhorn Approximation for Learning with Wasserstein Distance
• End-to-End Adversarial Text-to-Speech
• Spatio-Temporal Alignments: Optimal transport through space and time
• A Conditional Random Field for Discriminatively-trained Finite-state String Edit Distance

Cited by in corpus (3)

• Cycle-skipping mitigation using misfit measurements based on differentiable dynamic time warping
• Learning with Algorithmic Supervision via Continuous Relaxations
• Diffeomorphic Transformations for Time Series Analysis: An Efficient Approach to Nonlinear Warping