Anomaly Detection in BACnet systems
Meeting 32
Title: K-means testing
Date: 15/04/2020
Time: 6 pm
Location: Zoom Conference Call
Attendees: Sofian Ghazali, Rahul Balamurugan, Zahid Kamil
Absences: None
Action Items: Creating anomalous data for K-means, getting accuracy estimate
Meeting Agenda:
-
Create anomalous data
-
Run full K-means module
-
Test and get false-negative and false-positive scores
Discussion and Notes:
The meeting was started off by Rahul and Sofian rehashing the procedure for k-means detection. Zahid mentioned he made an anomalous data set, which was very welcome. After a brief explanation on the creation of the data set, Sofian immediately used it while screen-sharing so that we could help him in the testing. After debugging the code, the module passed the false-positive check, which was conducted by splitting the training and testing sets and using the randomly selected testing set to check if the anomaly detector read any of them as anomalies. For the false-negative or accuracy check, we used Zahid's created data set and found out the estimate of the accuracy.
Meeting Adjournment: 8:00 pm
Meeting Minutes prepared by: Rahul Balamurugan