Aims & Objectives

Download: Real-World Optimisation with Life-Long Learning background document (PDF)

Research Hypothesis

A hyper-heuristic framework which is able to continuously learn over time as it gathers and records experience learned from solving problems will be more efficient and effective at solving a range of practical problems than current optimisation techniques and will address current commercial concerns regarding the applicability of academic optimisation techniques to real-world problems.

Aims

  • to improve the current state of the art in generic problem-solvers by developing a novel system which exploits previous knowledge, continuously learns from experience and autonomously adapts to shifting problem characteristics
  • to demonstrate that the proposed system is more efficient and effective at producing high-quality solutions to real-world practical problems than previous optimisation approaches in terms of reducing costs and environmental impact
  • to develop an information database of problem-solving knowledge as a platform for advancing the development of optimisation techniques which are informed by real-world problems constraints and commercial priorities
  • to demonstrate to end-users and commercial software vendors that meta-heuristic optimisation techniques are applicable to the needs of the commerical world in terms of encapsulating real-world constraints and being cheap to both implement and maintain