Time-Series Representation Learning via Temporal and Contextual Contrasting

TL;DR

This paper introduces TS-TCC, an unsupervised framework for learning robust and discriminative representations from unlabeled time-series data using temporal and contextual contrasting modules, achieving competitive results with supervised methods.

Contribution

The paper proposes a novel unsupervised learning framework for time-series data that combines temporal and contextual contrasting modules to improve representation quality.

Findings

Linear classifiers on learned features perform comparably to supervised training.

TS-TCC is effective with few labeled data and in transfer learning scenarios.

The method outperforms existing unsupervised approaches on real-world datasets.

Abstract

Learning decent representations from unlabeled time-series data with temporal dynamics is a very challenging task. In this paper, we propose an unsupervised Time-Series representation learning framework via Temporal and Contextual Contrasting (TS-TCC), to learn time-series representation from unlabeled data. First, the raw time-series data are transformed into two different yet correlated views by using weak and strong augmentations. Second, we propose a novel temporal contrasting module to learn robust temporal representations by designing a tough cross-view prediction task. Last, to further learn discriminative representations, we propose a contextual contrasting module built upon the contexts from the temporal contrasting module. It attempts to maximize the similarity among different contexts of the same sample while minimizing similarity among contexts of different samples. Experiments have been carried out on three real-world time-series datasets. The results manifest that training a linear classifier on top of the features learned by our proposed TS-TCC performs comparably with the supervised training. Additionally, our proposed TS-TCC shows high efficiency in few-labeled data and transfer learning scenarios. The code is publicly available at https://github.com/emadeldeen24/TS-TCC.