I want to develop evolutionary algorithms that are as efficient as possible whilst being simultaneously as open-ended as possible. To this end I am currently working on using artificial chemistries as a way of constructing genetic systems that exhibit properties of gene regulation, gene expression and genome organisation, as part of my work on the Plazzmid project.
These questions lead me to speculate on what the computer science angle is on the origin of life. What were the properties of ancient chemical systems that kick-started the preservation of coded information that was used as a reference in subsequent chemistries?
It is important to recognise that not all problems are a good fit to these techniques - optimisation problems are not what I'm trying to solve here. Rather, I'm interested in applications where it is difficult or impossible to guess what the solution space might be. I'm interested in developing communities of organisms that continuously act to maintain their own fitness in response to changing conditions.
In summary then, I'm interested in the design principles of evolution. I believe that if we can understand these, and implement them in engineering systems, then we can use the vast processing power of current computational systems to their maximum potential.
Simon Hickinbotham (Research Associate) has been working on bioinspired computing since the early 1990s, and is now researching frameworks for open-ended evolution using artificial chemistries. His long term research goal is to develop a richer computational model of biological evolution that can be used to build engineering applications that are capable of developing innovative behaviour.
In 2008, he joined the department of Computer Science and YCCSA at the University of York. He works on the Plazzmid project within YCCSA, involving researchers in Biology, Electronics, and Computer Science. Other YCCSA work includes optimisation of artificial mutation networks (with Alastair Droop), artificial chemistries for both robot control (with Verena Fischer) and unconverional computing (with Fintan Nagle), and evolutionary optimisation of swarm model parameters (with Richard Coates)."
Between 1994 and 2008, he worked on a range of projects in computer vision and bioinspired computing. In reverese chronological order, these were:
- Raster to vector conversion of utility map images for the VISTA project, University of Leeds
- Web demonstrator for Auramol, for Cybyla Ltd and the University of York
- Detection of strain gauge failures using neural networks, University of York and BAe Systems
- Automated analysis of fish scale images, University of York and the Environment Agency