T-Rep: Representation Learning for Time Series using Time-Embeddings
This addresses the problem of handling complex time series data for machine learning practitioners, offering an incremental improvement over existing self-supervised methods.
The paper tackles the challenge of learning representations for multivariate time series, which are often unlabeled, high-dimensional, noisy, and contain missing data, by proposing T-Rep, a self-supervised method that learns time embeddings to extract temporal features and improve robustness; it outperforms existing self-supervised algorithms in classification, forecasting, and anomaly detection tasks, showing greater resilience in missing data regimes.
Multivariate time series present challenges to standard machine learning techniques, as they are often unlabeled, high dimensional, noisy, and contain missing data. To address this, we propose T-Rep, a self-supervised method to learn time series representations at a timestep granularity. T-Rep learns vector embeddings of time alongside its feature extractor, to extract temporal features such as trend, periodicity, or distribution shifts from the signal. These time-embeddings are leveraged in pretext tasks, to incorporate smooth and fine-grained temporal dependencies in the representations, as well as reinforce robustness to missing data. We evaluate T-Rep on downstream classification, forecasting, and anomaly detection tasks. It is compared to existing self-supervised algorithms for time series, which it outperforms in all three tasks. We test T-Rep in missing data regimes, where it proves more resilient than its counterparts. Finally, we provide latent space visualisation experiments, highlighting the interpretability of the learned representations.