Reproduce “Intrinsic Anomaly Detection for Multi-Variate 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

Stephan Rabanser, Tim Januschowski, Kashif Rasul, Oliver Borchert, Richard Kurle, Jan Gasthaus, Michael Bohlke-Schneider, Nicolas Papernot and 1 more

EnvInvEmb detects intrinsic anomalies by learning time-series embeddings that distinguish a broken environment–system relationship from unusual environmental conditions with an expected system response. It uses contrastive examples that break this relationship and adversarial training to discourage environmental information in the embedding. On Pendulum, it reaches 0.980 AUROC versus 0.951 for ResEmbRegr; residual thresholding performs better on DevOps and Turbine.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.