Evolutionary multiobjective optimisation: moving from methodologies to application

Most of the decisions we need to take in life require us to think about multiple different performance criteria simultaneously. For example, the buying decision on a desired product is rationally made in the pursuit of best price and quality possible. Often these criteria are in conflict, such that improvement in one (e.g. quality) cannot be achieved without detriment to the other (e.g. purchase budget). How can we support transparent, repeatable, decision-making in these circumstances?

A popular method to aid decision making is evolutionary multi-objective optimisation (EMO). By no longer focusing on a single target at a time, EMO techniques offer a more powerful view of the problem since several conflicting goals can be optimised in relation to each other.

In this talk, starting off with a brief introduction of evolutionary computation, I am going to talk about to what extent EMO can be enhanced by employing novel search structure, what structure is best for EMO to handle many objectives, and what EMO can really do for us in real life.