T-Rep: Representation Learning for Time Series using Time-Embeddings

arXiv:2310.04486v3 Announce Type: replace
Abstract: 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 …

abstract arxiv challenges cs.ai cs.lg data embeddings learn machine machine learning machine learning techniques multivariate representation representation learning series standard time series type vector vector embeddings