Reproduce “What Streaming Anomaly Detection Finds (and Misses) in Industrial 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
Magali Parrino, Antoine Ajenjo, Emmanuel Remy, Pierre Stephan, Paul Boniol
The authors evaluate anomaly detectors for filter-clogging monitoring at one nuclear-plant site, comparing methods that keep parameters fixed with methods that update as data arrives. They find that detector agreement marks prominent event clusters but often misses sparse events. A supervised top-k ensemble scores 0.353 VUS-PR, versus 0.346 for LODA; this single-site result does not establish clogging or prevent shutdowns.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.