Paper Accepted to Neverending Machine Learning '25

Our paper on identifying novel attack types at runtime has been accepted to NML. Congrats to Sumaita on her great work!

This project extends LOCoCAT and further analyzes CAN bus attacks on cars. Our framework uses statistical metrics to analyze detected attacks and determine whether they from a type the system has seen before or novel.

We evaluated the proposed framework using two datasets from the literature and found it can detect novel attacks with as little as one new data point!

Find more information about the paper here.