This centre-wide project investigates approaches to provide an understandable and integrated experience for people as they interact with AI systems. This research is aligned to the Understanding and enabling a smart data driven world theme. It includes knowledge representation, agent-based systems, knowledge-based systems, modelling dependencies, machine learning, clustering and aggregation
eXplainable AI (ARC Discovery 2021-2024 project) . To match performance of black-box systems with the benefit of explanation of the decisions taken. Employs fuzzy rule-based systems augmented with sophisticated co-dependent rules.
Human-aligned learning systems (Air Force Office of Scientific Research grant). Development of AI systems that operate autonomously in an environment that include other biologically-based intelligent agents. Emotion recognition/simulation, competency-awareness, privacy, security, natural language processing and generation, attribution, sensor interpretation, philosophical, cognitive science
Social networks modeling
Natural language processing
Advancing Optimization for Future Challenges
This multidisciplinary program addresses both the fundamentals of methods of optimisation and their applications in such fields as Defence, Transportation, Energy sector, Communications and Knowledge Engineering and Machine Learning. We develop advanced discrete and continuous nonlinear, nonsmooth, single and multiple objective optimisation models and algorithms, implement them on standard and high-performance hardware, including clusters, GPUs and quantum computing, and apply in the mentioned fields.
Includes the following projects
Fundamentals: Optimisation with incomplete knowledge, learning unknown objectives and parameters; Automatic modelling; Nonlinear Knapsack problems; Nonlinear multiextremal global optimisation; QUBO-based reformulations for quantum computing; Multibobjective optimisation; NLP, Knowledge Engineering and Modelling;
Defence: (DSTG funded projects) Whole-force design ; Nonlinear portfolio selection; Scheduling and timetabling for pilot training; Satcom downlink scheduling (DSTG)
Transportation: EV routing, charging, location and scheduling; Bus routing and scheduling (Hycel);
Energy and Industry: Digital twin – Energy microgrid; Preventive maintenance of solar assets
Knowledge Engineering and Machine Learning: Learning fuzzy measures and aggregation functions (ARC Discovery 2021-2024 project); Robust linear and nonlinear regression; Optimisation-based framework for Chebyshev approximation (ARC Discovery 2018-2020 project); Rational approximations and classifiers; Learning graph-based models; Scattered data approximation; Non-convex Clustering; Outliers detection; Learning Bayesian classifiers