Position: PhD student
I am working as a PhD student in the BDS group. The core focus of my research is the use of machine learning techniques to augment and improve the efficiency of Genetic Programming (GP) algorithms, a particular subfield of Evolutionary Computation. In general, I am interested in the intersection between the areas of machine learning, optimization and computational intelligence.
I have a BSc in Mathematics and Computing from UCC. Before joining the BDS group in 2013 to work as part of the SFI-funded Multi-Core Attributed Grammatical Evolution project, I was working for Disability Support Services in Cork Institute of Technology.
So far, my work for the BDS group has focused on two main research themes (both of which share the common goal of supporting and augmenting Genetic Programming).
Configuring a Genetic Program (GP), tailoring its search operators and various design settings to the particular problem being optimized, can, in practice, be a somewhat haphazard trial-and-error process. During this project, my goal is to develop more principled design methods for GP. This approach seeks to first better understand the problems being optimized, and then use this acquired understanding to guide the GP design process. Gaussian Random Functions (GRFs) are a popular family of machine learning models that have been successfully applied to the modelling of a wide range of difficult real world problems. I have been using GRF models to efficiently construct and design high-performance GP search operators.
I am also using Lamarckian learning/Memetic Algorithm approaches to improve GP's local search abilities. Such techniques can be relatively inexpensive for many typical GP problem domains. My research to date has used such approaches to counter code bloat, reduce premature convergence and augment GP's exploitative search capabilities.
Presentations MadeDate: April 23, 2015
Fergal Lane presented "The Principled Design of Genetic Programming Memetic Algorithms in a Gaussian Random Function Framework"
PhD transfer talk
Date: September 16, 2014
Fergal Lane presented "On Effective and Inexpensive Local Search Techniques in Genetic Programming Regression"
Parallel Problem Solving from Nature – PPSN XIII