There is plenty of scope for new technologies and theoretical results in the field of evolutionary dynamic optimisation, making it an exciting field of research and highly suitable for smaller research projects. We are happy to supervise students interested in any of the topics shown below. Alternatively, please feel free to contact us with your own suggestions if you have a particular concept in mind.

• New problem benchmark generators, either continuous or combinatorial.
  • The idea is to design new test-beds for researchers in the field to evaluate and compare their algorithms on. It would be important to argue for the design of the problem generator and references to real-world scenarios are encouraged. The design should also be accompanied by some (preliminary) analysis regarding the parameter settings a practitioner would have to choose (e.g., those controlling the dynamics). This type of project may involve a lot of coding and the code should be made publicly available upon completion of the project. Of particular interest are:
    • Multi-objective dynamic problem generators
    • Problem generators with dynamic constraints
    • Problem generators with time-linkage
      
• Design of new evolutionary algorithms for dynamic optimisation problems.
  • Traditional evolutionary algorithms need to be adjusted to cope well with dynamic optimisation problems. This work could be based on any existing EA or could be a new type of algorithm that combines different elements from existing technologies. The resulting algorithm would need to be tested and compared against existing approaches on any of the currently available benchmark problems. This type of project would involve algorithm design, coding and experimental work. Of particular interest would be the design of new estimation of distribution algorithms.
    
• Dynamic Graph Routing Problems
  • This project would be concerned with dynamic graph routing problems such as dynamic shortest path or dynamic minimum spanning tree. These are very important problems in both logistics and telecommunications and it may be possible to make use of some real-world data. The project would involve looking at different dynamic graph problems and proposing suitable evolutionary algorithms to produce high quality solutions for them.
    
• Experimental studies to advance our understanding of dynamic optimisation.
  • This project would investigate some theoretical aspects by empirical (or analytical) means. The results of these comparative studies are subsequently used to verify or contradict some of the commonly held assumptions in this field. These projects would require a fair amount of background reading and a fair amount of coding and experimental design. Example topics are as follows:
    • The impact of diversity on the performance of EAs in dynamic domains
    • The benefit of memory in cyclic, acyclic and random dynamics
    • Trade-off in dealing with the (stationary) base problem and the imposed dynamics
    • The problem of detecting change in the environment
    • Steady-state versus generational update schemes
    • The role of selection, crossover and mutation
    • Evaluation criteria for algorithms in the dynamic domain