Artificial Intelligence and Network Security: Trends and Challenges

Dear Colleagues,

Artificial intelligence (AI) technologies have shown great potential in many fields of network security, particularly in network traffic monitoring and in the detection of the unauthorised access, misuse or denial of networked resources.

However, well‑performing solutions presented in the current scientific literature often fail to generalize from the lab setting to real‑world deployments due to diverse operating conditions. In particular, state‑of‑the‑art AI models are designed assuming that data remain stationary, i.e., that the training data will remain representative of future test data. Accordingly, such models cannot naturally cope with the rapid evolution exhibited by network traffic patterns over time (also called “concept drift”) or with conveniently crafted attack data aimed to evade the detection system (Adversarial Machine Learning (AML) attacks). This problem is also exacerbated by the fact that such models provide non‑interpretable decisions, which also hinders their debugging process towards identifying whether they have learned spurious correlations present in the training dataset.

In addition, AI models for cybersecurity are often trained and evaluated assuming that the data (network traffic, system logs, etc.) are available in a single place. However, data handling in distributed application scenarios, especially when different administrative domains are involved, is not always an easy task, as such data might contain confidential information. 

This Special Issue is focused on novel research in AI-based algorithms, methods and architectures for the reliable detection of cyber threats (see the non-exhaustive list of topics below). More precisely, we encourage original works which have not been published or submitted for publication elsewhere, from both academia and industry, focussing on the challenges and trends related to concept drift,  AML attacks and model training with distributed data. This includes methods for explainability and error analysis of AI models that can be used to patch flaws in the training/testing pipeline. Survey papers that offer a perspective on related work and identify key challenges for future research will be considered as well. We look forward to your submissions!

Topics of interest for this Special Issue include, but are not limited to, the following:

• AI-based intrusion detection, tolerance, and prevention;
• Energy-efficient and resource-efficient AI-based intrusion detection;
• Algorithms for concept drift detection, understanding and adaptation in AI-based cyber-threat detection systems;
• Robustness of AI-based cybersecurity solutions to black-box, white-box, and grey-box adversarial attacks;
• Defences against training/testing attacks;
• Secure, privacy-preserving and federated machine learning in cybersecurity;
• AI-based cybersecurity solutions that are transparent by design;
• Detection of data bias and algorithmic bias in AI-based cybersecurity;
• Theoretical aspects of AI model interpretability in cybersecurity.

Dr. Roberto Doriguzzi-Corin
Dr. Vera Rimmer
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Mathematics is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2600 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.