Self-Supervised Contrastive Representation Learning for Time-Series Classification
Kunpeng Li, Jong‐Chul Lee · 2024
Time series lacks intuitive features, which increases the duration of annotation. Therefore, this paper proposes a self-supervised temporal contrastive model. By automatically extracting useful representations from unlabeled data, the model then uses a small amount of labeled data to achieve classification of time series. Extensive experiments demonstrate that the framework we propose possesses efficient and powerful learning capabilities.