Monday 3 November 2014, 1.00PM
Speaker(s): Dr Alastair Channon, Keele University
Artificial evolutionary systems often exhibit very low diversity and bounded complexity. I will give an overview of a number of projects, carried out with research partners and students, that include addressing these shortcomings within their aims: research into critical mutation rates for the maintenance of allelic diversity; mutation rate control for extinction-avoiding novel adaptation; the role of transcription errors in discovering behaviours inaccessible to incremental genetic evolution alone; and selection complexification strategies for incremental evolution. I will then look forward and discuss three approaches to the evolution of intelligent agents, more specifically to tackling the requirement to find an evolutionary path from a random or primordial soup to a sufficient behaviour: scaffolding selection with increasingly complex tasks, searching for novel behaviours, and seeding evolution by artificial selection with the results of (long-term or open-ended) evolution by natural selection. Examples of evolved agents will be shown in simple two-and three-dimensional environments.
Dr Alastair Channon worked in the software industry (at Micro Focus) before carrying out his BA/MA in Mathematics at the University of Cambridge and then focussing on Evolutionary and Adaptive Computation through an MSc at the University of Sussex and a PhD at the University of Southampton. From there he moved directly to a senior lectureship (post-92) at the University of Portsmouth in 1999, a lectureship at the University of Birmingham in 2004 and to the School of Computing and Mathematics at Keele University in 2007, where he is now a senior lecturer. His primary research interest is in the open-ended evolution of neurally controlled animats and he is best known for having created the only closed system other than Earth's biosphere to have passed the enhanced statistical "ALife Test" for open-ended evolution. Alastair's recent publications have included significant results on the relationship of mutation rate to population size, with clear implications for biological extinction events, and to fitness, computed over both abstract and biological fitness landscapes. He is a partner in a BBSRC project on the theory and practice of evolvability: effects and mechanisms of mutation rate plasticity, with partners at Manchester and Middlesex Universities, following the same team's successful completion of an EPSRC project on information dynamics in evolutionary systems.