The complexity of current systems and services entails an increase in the volume and variety of information available, as well as the risks associated with it. From this perspective, the use of automatic modeling and learning schemes is appropriate in order to improve the protection and security of systems and services.

In this area, Artificial Intelligence (AI) is called upon to provide advantages from different perspectives

• Greater generalizability and adaptability to new threats and attacks (zero-day attacks).
• Reduction of positive / negative faults in the case of detection.
• Distributed learning modeling based on federated learning.
• Ability to integrate threat intelligence from different sources derived from common knowledge.
• Generation of improved training data sets resulting in robust machine learning models.
• Automatic assessment of the security risk and exposure of an entity.
• Automation of the decision-making process and response to attacks or security threats.

The NESG group investigates and develops new applied AI solutions for the provision and modeling of robust systems from the security point of view through the use of optimization techniques, machine learning, deep machine learning and reinforcement learning. Likewise, the previous models are also susceptible to being attacked in all phases of their development. Consequently, the NESG studies and addresses the creation of secure machine learning models (security-by-design) in the face of, for example, the intentional generation of samples that lead to erroneous and, on the other hand, intentional decision making.