Unsupervised Scalable Representation Learning for Multivariate Time Series
This addresses the problem of handling variable-length and sparsely labeled time series for machine learning practitioners, though it appears incremental as it builds on existing representation learning approaches.
The paper tackles the challenge of learning universal embeddings for multivariate time series by proposing an unsupervised method that combines causal dilated convolutions with a novel triplet loss, demonstrating scalability and quality in experiments.
Time series constitute a challenging data type for machine learning algorithms, due to their highly variable lengths and sparse labeling in practice. In this paper, we tackle this challenge by proposing an unsupervised method to learn universal embeddings of time series. Unlike previous works, it is scalable with respect to their length and we demonstrate the quality, transferability and practicability of the learned representations with thorough experiments and comparisons. To this end, we combine an encoder based on causal dilated convolutions with a novel triplet loss employing time-based negative sampling, obtaining general-purpose representations for variable length and multivariate time series.