Core Methods and Applications of Artificial Intelligence (CORE-AI)

CORE-AI advances foundational and translational Artificial Intelligence methods that enable understanding, simulation, and intervention in complex real-world systems. The group develops and integrates:

• explainable and causal AI
• hybrid AI-driven modelling
• structured and graph-based machine learning
• transformer-based and foundation model architectures
• large language models
• federated and distributed learning frameworks

Our work moves beyond purely predictive paradigms towards AI systems that support causality, explainability, counterfactual analysis, and decision-making under uncertainty. We are particularly interested in methodological frameworks that combine theoretical robustness, computational scalability, and federated and distributed learning paradigms to enable AI systems that learn effectively across heterogeneous data sources and dynamic environments while remaining deployable in real-world settings.

A central objective of CORE-AI is to translate methodological advances into system-level tools, including:

• digital twins
• AI virtual assistants
• decision-support systems

These tools are designed for domains characterised by complexity, temporal dynamics, partial observability, and heterogeneous or data-constrained environments.