software · Zenodo (CERN European Organization for Nuclear Research)
This release provides a comprehensive reproducibility package designed to support a multi-environment transport audit of industrial anomaly detection models. It contains the locked development benchmarks, calibration records, confirmatory score streams, event ledgers, and diagnostic analyses used to scrutinise whether model success on standard benchmarks translates to reliable operational assurance. The dataset preserves evaluation materials covering held-out environments, including HAI 22.04, Sherlock Basic, and Sherlock Semiurban, while tracking specific decisions such as the PCAResidual method. It incorporates verification reports, source code, and correction manifests, specifically noting where earlier event-derived outputs for Sherlock Basic have been superseded by a corrected evaluation of eighteen attacks. Access to the underlying raw industrial datasets must be obtained separately from their primary sources.
Industrial systems rely on anomaly detectors to prevent failures and detect cyber attacks, but high scores in laboratory tests do not guarantee reliable performance in operation. Providing open verification tools and data audits helps researchers and operators confirm whether safety-critical monitoring systems function properly across different industrial environments.
The materials can be used by industrial automation engineers, security auditors, and system integrators seeking to independently verify or test anomaly detection tools prior to plant deployment. Because this artifact serves as an analytical audit and benchmarking suite rather than a standalone commercial product, it functions at the validation and testing stage of the technology development pipeline.
AI-generated from the published abstract. Always read the original work before citing.
This reproducibility artifact contains the frozen development benchmark lock, confirmatory score streams and calibration records, event ledgers, correction manifests, post-hoc diagnostic analyses, source code, and verification reports supporting a multi-environment audit of industrial anomaly detectors. The artifact preserves the development-supported PCAResidual decision and the held-out HAI 22.04, Sherlock Basic, and Sherlock Semiurban evidence used to evaluate deployment-facing assurance screening. Raw third-party SWaT, HAI, and Sherlock datasets are not redistributed; users must obtain them from their official sources under the applicable licences. The Sherlock Basic directory preserves both the original frozen-score evidence and the corrected official 18-attack event evaluation, with the original event-derived outputs explicitly marked as superseded for scientific inference.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.5281/zenodo.22067730
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
