Hands-On Artificial Intelligence for Cybersecurity

Empowering detection systems with human feedback

From what we have seen so far, the best use of anomaly detection systems sees the interaction of automated procedures with the specialized activities carried out by human operators.

Therefore, the use of the anomaly detection systems as a support tool for human specialists allows the mitigation of the costs deriving from false positives, at the same time improving the ability to reduce false negatives by exploiting human feedback (as in the aforementioned case of improving the reliability of the classification sample labels used to train supervised algorithms).

However, this human-machine synergy presupposes that algorithms are less opaque and more easily interpreted by humans, therefore ...