Multi-Objective Neuroevolution for Digital Healthcare Applications

Abstract

In digital healthcare, the problems are inherently noisy and multi-objective involving the trade-off between several objectives. Evolutionary Algorithms are stochastic optimisation methods which take inspiration from the natural (or biological) evolutionary principles for constituting specific optimisation and search procedures. They can produce a set of trade-off solutions for consideration with desirable characteristics (e.g. diversity, pertinence, proximity etc.) in a multi-objective set. Multi-objective Evolutionary Algorithms were exploited for the training stage of machine learning algorithms. The research mainly aims to investigate the potential of applicability of Neuroevolution to the digital healthcare industry. The precision of the digital health measurements has vital importance, and it has been hypothesised that with advanced machine learning techniques it is possible to revolutionise this field from the multi-objective perspective.

Researchers