Enzyme Genetic Programming: Modelling Biological Evolvability in Genetic Programming. Lones, M. A. Ph.D. Thesis, The University of York, Heslington, York, YO10 5DD, UK, sep, 2003.
Enzyme Genetic Programming: Modelling Biological Evolvability in Genetic Programming [link]Paper  abstract   bibtex   
This thesis introduces a new approach to program representation in genetic programming in which interactions between program components are expressed in terms of a component's behaviour rather through its relative position within a representation or through other non-behavioural systems of reference. This approach has the advantage that a component's behaviour is expressed in a way that is independent of any particular program it finds itself within; and thereby overcomes the problem when using conventional program representations whereby program components lose their behavioural context following recombination. More generally, this implicit context representation leads to a process of meaningful variation filtering; whereby inappropriate change induced by variation operators can be wholly or partially ignored. This occurs as a consequence of program behaviours emerging from the self-organisation of program components, ignoring those components which do not fit the contexts declared by the other components within the program. This process results in gradual change within the behaviour of a program during evolution. This thesis also presents results which show that implicit context representation leads to better size evolution characteristics than conventional genetic programming; and that functional redundancy and Lamarckian reinforcement learning both improve evolutionary search, agreeing with previous research by other authors.
@PhdThesis{lones2003thesis,
  author =       {Michael A. Lones},
  title =        {Enzyme Genetic Programming: Modelling Biological
                 Evolvability in Genetic Programming},
  school =       {The University of York},
  year =         {2003},
  address =      {Heslington, York, YO10 5DD, UK},
  month =        {sep},
  keywords =     {genetic algorithms, genetic programming, evolvability,
                 representation, self-organisation, biological
                 modelling},
  URL =          {http://www-users.york.ac.uk/~mal503/common/thesis/main.html},
  size =         {200 pages},
  abstract =     {This thesis introduces a new approach to program
                 representation in genetic programming in which
                 interactions between program components are expressed
                 in terms of a component's behaviour rather through its
                 relative position within a representation or through
                 other non-behavioural systems of reference. This
                 approach has the advantage that a component's behaviour
                 is expressed in a way that is independent of any
                 particular program it finds itself within; and thereby
                 overcomes the problem when using conventional program
                 representations whereby program components lose their
                 behavioural context following recombination. More
                 generally, this implicit context representation leads
                 to a process of meaningful variation filtering; whereby
                 inappropriate change induced by variation operators can
                 be wholly or partially ignored. This occurs as a
                 consequence of program behaviours emerging from the
                 self-organisation of program components, ignoring those
                 components which do not fit the contexts declared by
                 the other components within the program. This process
                 results in gradual change within the behaviour of a
                 program during evolution. This thesis also presents
                 results which show that implicit context representation
                 leads to better size evolution characteristics than
                 conventional genetic programming; and that functional
                 redundancy and Lamarckian reinforcement learning both
                 improve evolutionary search, agreeing with previous
                 research by other authors.},
}

Downloads: 0