<p>This dataset contains manipulated outputs from the Tennesse Eastman Process (TEP), an anonymized chemical process. </p>
<p><br></p>
<p>To produce each manipulated output, the value of a single TEP feature modified for two hours. The subsequent TEP response is recorded. </p>
<p><br></p>
<p>In total, 286 manipulations are performed and recorded: each manipulation varies in (i) its magnitude, (ii) the pattern of the manipulation, and (iii) which TEP feature was manipulated. </p>
Funding
This work is supported in part by the Secure and Private IoT initiative at Carnegie Mellon CyLab (IoT@CyLab) and by Mitsubishi Heavy Industries through the Carnegie Mellon CyLab partnership program; by the U.S. Army Research Office and the U.S. Army Futures Command under contract W911NF-20-D-0002; by DARPA under contract HR00112020006; and by the U.S. Department of Defense under contract FA8702-15-D-0002.
Attributions for ML-based ICS Anomaly Detection: From Theory to Practice.
Clement Fung, Eric Zeng, and Lujo Bauer.
In Proceedings of the 31st Network and Distributed System Security Symposium, February 2024.