Adaptive Multi-Objective Search in Expensive High-Dimensional Socio-Technical Systems

This PhD project focuses on developing advanced machine learning techniques for solving optimisation problems in complex socio-technical systems, in collaboration with Honda Research Institute Europe. Modern systems such as electric vehicle (EV) energy networks involve large-scale decision-making under multiple competing objectives, including efficiency, fairness, sustainability, and user satisfaction. These problems are computationally expensive and high-dimensional, making traditional optimisation approaches insufficient. The project will explore cutting-edge approaches such as: Bayesian optimisation for efficient search under limited evaluation budgets Gaussian processes for surrogate modelling of expensive objective functions Multi-objective optimisation to balance competing criteria Adaptive variable selection for navigating high-dimensional spaces Key research goals include: Efficient exploration of large decision spaces using adaptive search strategies Modelling trade-offs between conflicting objectives using multi-output models Incorporating human-centred criteria such as trust, fairness, and explainability Developing scalable optimisation methods for real-world industrial systems The successful candidate will benefit from close collaboration with industry, including research visits and engagement with Honda’s international research community.

Strong background in machine learning, statistics, applied mathematics, or optimisation
Experience with probabilistic modelling or optimisation methods
Knowledge of Bayesian optimisation or Gaussian processes (desirable)
Strong programming and analytical skills

Apply via the University of Manchester application portal under PhD in Artificial Intelligence (AI CDT)
Required documents:
CV
Academic transcripts and certificates
Supporting statement (1–2 pages)
Two referees
English language certificate (if applicable)
Ensure you:
Mention the project title
Include supervisor names
Submit all documents before the deadline