Genetic Programming: Projects
The group carries out a variety of research in the field of genetic programming, with particular focus upon the Cartesian Genetic Programming method and upon the design of evolvable, scalable, biologically-motivated representations for genetic programming. The latter includes work upon developmental representations, implicit context representations, and multiple chromosome approaches. In addition to our theoretical work, we also have experience of applying genetic programming methods to a variety of real world problems, including robotic control, biological sequence understanding, medical classification, and image processing.
 |
AlBiNo - Artificial Biochemical Networks: Computational Models and ArchitecturesMembers: Michael Lones, Luis Fuente, Alex Turner, Andy Tyrrell, Susan Stepney (CS), Leo Caves (Biology) Biochemical networks are one of the most complex sets of structures found in biological systems. These structures are fundamental to the development, function and evolution of biological organisms, and are the main factor underlying the complexity seen within higher organisms. This 5 year EPSRC-funded project (grant no: EP/F060041/1) will promote the development and understanding of artificial biochemical network models and show how how they may be applied to complex real world computational tasks. |
 |
Biosequence Motif Discovery using Evolutionary ComputationMembers: Michael Lones, Andy Tyrrell We are looking at how evolutionary computation approaches can be used to discover patterns in biological sequence data. Our current focus is on finding regulatory motifs in DNA promoter regions, for which we have recently developed a population clustering evolutionary algorithm capable of finding multiple strong and weak signals within relatively long promoter sequences. We are also looking at at the use of co-evolution to improve search and characterise the structure of regulatory regions. This work is in collaboration with Prof. Finn Drabløs in the Department of Cancer Research and Molecular Medicine at NTNU. |
 |
Cartesian Genetic Programming (CGP)Members: Julian Miller, Steve Smith, David Halliday, Andy Tyrrell, Tim Clarke, Janet Clegg, Michael Lones, James Alfred Walker, Martin Trefzer, James Hilder, Tuze Kuyucu, Gul Muhammad Khan, Katharina Voelk Cartesian Genetic Programming (CGP) is a form of Genetic Programming that encodes a directed graph representation of a computer program. It was invented by Julian Miller in 1998 and was developed from a representation for evolving electronic circuits devised by Julian Miller and Peter Thomson a few years earlier. The Intelligent Systems Group at the University of York is currently developing a number of extensions to CGP which improve the performance and evolvability of the original technique. Also, members of the Intelligent Systems Group are actively applying CGP to a number of real-world problems and applications. |
 |
Implicit Context Representation for Evolutionary ComputationMembers: Michael Lones, Andy Tyrrell, Steve Smith Implicit context is a novel approach to representing evolving structures in which structural components describe their interactions using functional descriptions of other components. This results in positional independence within the structure's evolving representation, producing better performance when recombination operators are used. |
 |
Molecular Software and Hardware for Programmed Chemical SynthesisMembers: Andy M. Tyrrell, Cristina Costa Santini Synthetic Ribosomes - Software-Controlled Assembly of Oligomers: This is a very interdisciplinary project. We intend to design a nanoscale chemical factory in which the machines, like the products, are molecules. The factory will not only build molecules but will be capable of evolving them to have desirable properties. The products will be linear molecules produced by linking together smaller building blocks in a defined sequence - at each stage the molecular machinery will be capable of choosing the correct building block from a range of possibilities. The system will be capable of synthesizing a library of molecules with different sequences and selecting 'successful' molecules for their fitness to perform a specified task. We will also develop designs for more powerful systems in which the molecular machinery responsible for chemical synthesis has internal computing power and can direct its own operation. |