Towards Enhancing Time Series Contrastive Learning: A Dynamic Bad Pair Mining Approach

TL;DR

This paper introduces a Dynamic Bad Pair Mining algorithm to identify and suppress harmful positive pairs in time series contrastive learning, improving representation quality by reducing noise and faulty alignments.

Contribution

The paper proposes a lightweight, parameter-free method that dynamically detects and down-weights bad positive pairs, enhancing existing contrastive learning techniques for time series data.

Findings

DBPM effectively reduces the impact of noisy and faulty positive pairs.

Experiments show improved representation quality on four large-scale datasets.

The method is compatible as a plug-in with current state-of-the-art models.

Abstract

Not all positive pairs are beneficial to time series contrastive learning. In this paper, we study two types of bad positive pairs that can impair the quality of time series representation learned through contrastive learning: the noisy positive pair and the faulty positive pair. We observe that, with the presence of noisy positive pairs, the model tends to simply learn the pattern of noise (Noisy Alignment). Meanwhile, when faulty positive pairs arise, the model wastes considerable amount of effort aligning non-representative patterns (Faulty Alignment). To address this problem, we propose a Dynamic Bad Pair Mining (DBPM) algorithm, which reliably identifies and suppresses bad positive pairs in time series contrastive learning. Specifically, DBPM utilizes a memory module to dynamically track the training behavior of each positive pair along training process. This allows us to identify potential bad positive pairs at each epoch based on their historical training behaviors. The identified bad pairs are subsequently down-weighted through a transformation module, thereby mitigating their negative impact on the representation learning process. DBPM is a simple algorithm designed as a lightweight plug-in without learnable parameters to enhance the performance of existing state-of-the-art methods. Through extensive experiments conducted on four large-scale, real-world time series datasets, we demonstrate DBPM's efficacy in mitigating the adverse effects of bad positive pairs.