Hands-On Artificial Intelligence for Cybersecurity

Anomaly-driven IDS

With the introduction of AI techniques to the field of NIDS, it is now possible to evolve traditional IDS toward more advanced detection solutions, exploiting supervised and unsupervised learning algorithms, as well as reinforcement learning and deep learning.

Similarly, the clustering techniques analyzed in the previous chapters, which exploit the concepts of similarity between the categories of data, can validly be used for the implementation of anomaly-based IDS.

In choosing the algorithms for the anomaly detection network, however, some characteristic aspects of the network environment must be taken into consideration:

• In the case of solutions based on supervised learning algorithms, we will have to categorize (label) ...