S2RAI
Safe, Secure and Robust AI Track at the SAC26 Conference
2025/06/27: Track website is OUT!
Machine Learning (ML) became the enabling technology of several applications. There is an enormous interest in adopting them to support critical tasks such as error/intrusion/anomaly detection, image classification or even object detection and trajectory planning to support autonomous driving. However, components that operate in critical systems must be assessed so that the encompassing system complies with adequate safety and security requirements before they are deployed and used in their operational environment without generating catastrophic hazards to the health of citizens, the environment, or the infrastructures. The absence of safety/security guarantees, or even a quantitative estimation of their robustness against specific events (e.g., unknown or zero-day attacks, adversarial attacks, out-of-distribution data) may be considered unacceptable by certification bodies, holding their deployment into real critical systems.
Unfortunately, the current maturity of research is still far from guaranteeing a safe, secure and robust operation of ML algorithms in many domains. Therefore, this track calls for contributions that aim at making a step towards a safe, secure and robust operation of ML components through (non-exhaustive list):
the definition of architectural patterns (e.g., safety monitors or safety wrappers), software mechanisms, model-based analyses, and metrics aiming at a safe operation of ML algorithms;
the proposal or experimental evaluation of defenses against adversarial attacks or any potential threat to ML algorithms themselves;
techniques to make ML algorithms robust against unknown, out-of-distribution data, and their evaluation in simulated environments or in the wild;
methods for explainable AI, their evaluation, comparison and applicability to critical systems;
frameworks or mechanisms to precisely compute the confidence in the prediction of ML algorithms, and use them to build a safe, secure or robust ML-based component;
methodologies or data-driven processes that aim at guiding the integration of ML algorithms in critical systems such that safety and security requirements of the encompassing systems can be guaranteed.
September 26, 2025: Submission of regular papers and SRC research abstracts
October 31, 2025: Notification of paper/SRC acceptance/rejection
December 5, 2025: Camera-ready copies of accepted papers/SRC
December 12, 2025: Author registration due date
March 23 - 27, 2026: Conference in Thessaloniki, Greece
SAC (and thus S2RAI) accepts to types of submissions: regular papers and SRC Abstracts. The review process is double-blind: please prepare your paper accordingly. You can download the official Autor Kit from here (contains MS Word and Latex-Overleaf templates).
Regular papers: the total number of pages allowed with your registration without any extra page fee is eight (8). See the SAC 2025 website registration page for additional page charges. A maximum of 10 (first 8 pages no extra fees, 2 extra pages with fees) is allowed for each regular paper.
Student Research Track (SRC) Abstracts: the total number of pages allowed with your registration is four (4). No extra pages. The conference provides travel grants for PhD students, check here for further information.
Note that those page limits include references.
Regular papers: see here
SRC papers: see here
Paper registration is required, allowing the inclusion of the paper/poster in the conference proceedings. An author or a proxy attending SAC MUST present the paper paying a full conference fee. This is a requirement for the paper/poster to be included in the ACM digital library. No-show of registered papers and posters will result in excluding them from the ACM digital library.
Descending surname ordering
Simon Burton (University of York, UK)
Marta Catillo (University of Sannio, Italy)
Angelo Sotgiu (University of Cagliari, Italy)
Jeremie Guiochet (LAAS CNRS, University of Toulouse, France)
Tommaso Zoppi (University of Florence, Italy)
Should you have any questions, please mail tommaso<dot>zoppi<at>unifi<dot>it
TBA