cfeng783 / ndss19_invariantrulead Goto Github PK

I've been reading your 2019 article, "A Systematic Framework to Generate Invariants for Anomaly Detection in Industrial Control Systems“. It mentioned that you’ve used the WADI data set. In The data set description section in this paper, you pointed out that "The data is collected every second by running WADI non-stop for a total of 16 days. Each data point consists of 103 attributes, among which 67 are continuous sensor readings and the remaining are discrete actuator states."

However, the data set I downloaded from the WADI website shows that WADI has a total of 127 dimensions, of which the sensor part has 67 dimensions. So I would like to ask which actuators were deleted when you mentioned that you used a total of 103 dimension features.

Thank you very much!