Unsupervised Representation Learning for Time Series with Temporal Neighborhood Coding
This work addresses the challenge of modeling complex time series in domains like medicine where labeling is impractical, offering an incremental improvement over existing unsupervised methods.
The paper tackles the problem of learning generalizable representations for non-stationary time series with sparse labels by proposing Temporal Neighborhood Coding (TNC), a self-supervised framework that uses a debiased contrastive objective to distinguish signals within temporal neighborhoods from non-neighboring ones, resulting in superior performance on clustering and classification tasks across multiple datasets.
Time series are often complex and rich in information but sparsely labeled and therefore challenging to model. In this paper, we propose a self-supervised framework for learning generalizable representations for non-stationary time series. Our approach, called Temporal Neighborhood Coding (TNC), takes advantage of the local smoothness of a signal's generative process to define neighborhoods in time with stationary properties. Using a debiased contrastive objective, our framework learns time series representations by ensuring that in the encoding space, the distribution of signals from within a neighborhood is distinguishable from the distribution of non-neighboring signals. Our motivation stems from the medical field, where the ability to model the dynamic nature of time series data is especially valuable for identifying, tracking, and predicting the underlying patients' latent states in settings where labeling data is practically impossible. We compare our method to recently developed unsupervised representation learning approaches and demonstrate superior performance on clustering and classification tasks for multiple datasets.