Reproduce “Transparent Networks for Multivariate Time Series” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Minkyu Kim, Suan Lee, Jinho Kim
GATSM models multivariate time series with additive feature contributions and learned temporal weighting. It transforms features with shared neural basis functions and applies masked attention over time. Across eight real-world tasks, it has the best reported average rank; on Heartbeat, it reaches 0.843 AUROC versus 0.724 for NATM and 0.690 for Transformer, but does not lead every task.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.