MICOS: Mixed supervised contrastive learning for multivariate time series classification

Multivariate time series (MTS) classification is an emerging field with increasing demand. Existing representation learning methods for MTS classification are generally based on self-supervised learning. This results in their inability to maximize the use of labels. In addition, the inherent complexity of MTS makes it difficult to learn latent features. To this end, we introduce a new Mixed Supervised Contrastive Loss (MSCL) for MTS representation learning. To effectively leverage labels, the MSCL is calculated by mixing self-supervised, intra-class and inter-class supervised contrastive learning approaches. Then, based on MSCL, we further propose a novel MIxed supervised COntrastive learning framework for MTS classification (MICOS). It uses the spatial and temporal channels to extract the complicated spatio-temporal features of MTS. Additionally, the MSCL is applied at the timestamp level to capture the multiscale contextual information. Experiments were carried out by performing supervised and self-supervised classification tasks on 30 public datasets from the UEA MTS archives. The results show the reliability and efficiency of MICOS compared with 13 competitive baselines.(c) 2022 Elsevier B.V. All rights reserved.

Automated summary

Multivariate time series (MTS) classification is a growing field with increasing demand. Existing representation learning methods for MTS classification are limited in utilizing labels due to their reliance on self-supervised learning. To address this, a new Mixed Supervised Contrastive Loss (MSCL) is introduced for MTS representation learning, which combines self-supervised, intra-class, and inter-class supervised contrastive learning approaches. Based on MSCL, a novel Mixed supervised Contrastive learning framework for MTS classification (MICOS) is proposed, utilizing spatial and temporal channels to extract complex spatio-temporal features and applying MSCL at the timestamp level to capture multiscale contextual information. Experimental results on 30 public datasets from the UEA MTS archives demonstrate the reliability and efficiency of MICOS compared to 13 competitive baselines.