Reproduce “Recurrent Neural Networks for Multivariate Time Series with Missing Values” 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
Zhengping Che, Sanjay Purushotham, Kyunghyun Cho, David Sontag, Yan Liu
GRU-D predicts outcomes from irregular time-series by using measurements, masks, and time intervals. It learns to decay stale values toward variable means and reduce hidden-state information as time passes. In retrospective MIMIC-III in-hospital mortality experiments, its AUC was 0.8527 ± 0.003, versus 0.8380 ± 0.008 for GRU-simple. It matched a baseline with 36 rather than 48 hours of input, not proving benefit.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.