A Unified Contrastive-Generative Framework for Time Series Classification

TL;DR

This paper introduces CoGenT, a novel framework that unifies contrastive and generative self-supervised learning for time series classification, improving robustness and accuracy across diverse datasets.

Contribution

It is the first to combine contrastive and generative SSL paradigms for time series, addressing their individual limitations through joint optimization.

Findings

Achieves up to 59.2% F1 improvement over SimCLR.

Achieves up to 14.27% F1 improvement over MAE.

Demonstrates consistent performance gains across six datasets.

Abstract

Self-supervised learning (SSL) for multivariate time series mainly includes two paradigms: contrastive methods that excel at instance discrimination and generative approaches that model data distributions. While effective individually, their complementary potential remains unexplored. We propose a Contrastive Generative Time series framework (CoGenT), the first framework to unify these paradigms through joint contrastive-generative optimization. CoGenT addresses fundamental limitations of both approaches: it overcomes contrastive learning's sensitivity to high intra-class similarity in temporal data while reducing generative methods' dependence on large datasets. We evaluate CoGenT on six diverse time series datasets. The results show consistent improvements, with up to 59.2% and 14.27% F1 gains over standalone SimCLR and MAE, respectively. Our analysis reveals that the hybrid objective preserves discriminative power while acquiring generative robustness. These findings establish a foundation for hybrid SSL in temporal domains. We will release the code shortly.