A Co-training Approach for Noisy Time Series Learning

TL;DR

This paper introduces TS-CoT, a co-training contrastive learning method that enhances noisy time series representation learning by leveraging complementary views, leading to improved performance and transferability.

Contribution

The paper proposes a novel co-training approach for noisy time series learning that uses two different encoders to create complementary views, improving robustness and accuracy.

Findings

TS-CoT outperforms existing methods on four benchmarks.

The approach effectively mitigates data noise and corruption.

Learned representations transfer well to downstream tasks.

Abstract

In this work, we focus on robust time series representation learning. Our assumption is that real-world time series is noisy and complementary information from different views of the same time series plays an important role while analyzing noisy input. Based on this, we create two views for the input time series through two different encoders. We conduct co-training based contrastive learning iteratively to learn the encoders. Our experiments demonstrate that this co-training approach leads to a significant improvement in performance. Especially, by leveraging the complementary information from different views, our proposed TS-CoT method can mitigate the impact of data noise and corruption. Empirical evaluations on four time series benchmarks in unsupervised and semi-supervised settings reveal that TS-CoT outperforms existing methods. Furthermore, the representations learned by TS-CoT can transfer well to downstream tasks through fine-tuning.