Properly working critical infrastructures such as those delivering electricity, water, transportation, and gas are essential for our modern society. Proper and safe operation of such infrastructure requires dealing with many aspects and challenges such as: maintaining stability and functionality of the entire system, anomaly detection, defense against cyber-physical attacks, etc. Usually, the human operator monitors and controls the system by investigating (in real-time) sensor readings from the system. With recent advances in AI (in particular in Machine Learning), the need to design data-driven decision support systems for critical infrastructure arises. However, such data-driven systems have to deal with specific and unique challenges such as complex spatio-temporal data (e.g. sensor readings), uncertainty and noise and the input data as well as in the topology and description of the system, very sparse sensor placements, missing or uncertain labels, simulated training data only, etc. Data-driven systems applied to critical infrastructure have been classified as high-risk systems (e.g. by the EU AI act) because failures can have catastrophic consequences in the real world. Consequently, such systems must (by law, such as requested by the EU AI act) provide transparent outputs, and also come with guarantees with respect to robustness.
In this special session, we aim to bring together researchers from AI and engineering, as well as practitioners from industry, to explore and discuss foundational research and applications of data-driven methods in critical infrastructure. We expect high-quality and novel contributions to this highly relevant topic.
We welcome contributions to all topics related to Machine Learning in critical infrastructure, including (but not limited to):
Transparency of data-driven systems in critical infrastructure
Handling noise and uncertainty in data-driven systems
Data-driven systems for spatio-temporal data
Robustness of data-driven systems
Novel applications of machine learning to critical infrastructure
Submissions must be anonymized and submitted directly through the EDAS system (https://edas.info/newPaper.php?c=31628&track=121739) – make sure to select the topic “Special Session: Machine Learning in Critical Infrastructure”. The complete author guidelines can be found at https://2024.ieeewcci.org/submission
Submission deadline: 15.01.2024 29.01.2024 (extended)
André Artelt, Bielefeld University, Germany
Prof. Cesare Alippi, Politecnico di Milano, Italy & Università della Svizzera italiana, Switzerlan
Prof. Barbara Hammer, Bielefeld University, Germany
Prof. Marios M. Polycarpou, University of Cyprus, Cyprus