James Butterworth

Computer Scientist

Research

My research interests lie mainly in the area of Genetic Algorithms and Neuroevolution. Given that evolution is the only process that we know of to produce the complex intelligence on Earth, my thoughts wander around variants of the question: ‘Why do current evolutionary methods not result in the sort of complex intelligence seen in nature?‘. I do not deny that there may be other methods such as Gradient Descent that can produce this type of intelligence, but that still does not account for why our current genetic algorithms are failing to live up to our high expectations.

The ways in which our current evolutionary algorithms differ from the processes that occur in nature are likely to be plentiful and could be based on our genotypical or phenotypical representations, or our evolutionary processes: genetic operators and our optimisation procedures. The reason for the disparity between the products of natural and simulated evolution is still a very large open question.

Currently I find myself focusing in on one of these areas, the genotypical and phenotypical representations of Neuroevolutionary structures and the mapping/developmental process that occurs between them.

Additionally, I have been involved in a number of projects that have seen the realisation of some Bio-Inspired optimisation techniques on real robots - specifically Micro-Aerial Vehicles.