Reproduce “Explainable Online Unsupervised Anomaly Detection for Cyber-Physical Systems via Causal Discovery from 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
Daniele Meli
The authors present an anomaly detector that learns temporal causal relationships among variables during normal operation, then flags links whose coefficients change beyond link-specific thresholds. They evaluate it on SWAT and Pepper, using altered links to rank affected variables. It detects all anomaly points in SWAT scenarios, while Pepper recall is 92%, below T-C-E’s 95%; results are limited to these tests.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.