[Re] Unsupervised Scalable Representation Learning for Multivariate Time Series

Felix Liljefors*, Mohammad Moein Sorkhei*, Sofia Broomé*

Abstract: We have reimplemented the paper Unsupervised Scalable Representation Learning for Multivariate Time Series by Franceschi et al., as part of the Replications track of the NeurIPS Reproducibility Challenge 2019. Among a large number of experiments, we were largely able to replicate the original paper's results. Since our experiments were run with a set of additional assumptions, this speaks in favor of the robustness of the authors' method.

Track: Replicability

NeurIPS Paper Id: https://openreview.net/forum?id=ryloJjA4d4