Edinburgh Napier University

  Optimisation – finding cost-effective or high-performing solutions - is a key economic driver for business today. However, academic literature on search-based optimisation techniques reflects an escalating arms race to produce highly specialized algorithms that are too complex for business to apply. In contrast, the machine-learning community has made huge advances in developing ensemble-methods that derive their power from aggregating many weak classifiers, resulting in versatile, easy-to-use algorithms. This research proposes to lay the theoretical basis for a paradigm shift in optimisation, exploiting existing theory from Machine-Learning to create ensembles-ofoptimisers
that are both effective and applicable in industry.