A biologically based approach to the mutation of code

• Main Author: Vandenberg, Loretta L.
• Other Authors: Kidd, Taylor
• Language: en_US
• Published: Monterey, California. Naval Postgraduate School 2012
• Online Access: http://hdl.handle.net/10945/8032

Approved for public release; distribution is unlimited === Evolutionary programming is a relatively new problem solving approach in the field of computer science. It attempts to model the processes of natural selection and evolution to solve complex problems. This technique is very powerful because it can be applied to a wide range of problems, and can find solutions that other more traditional techniques cannot. This research attempts to augment the methodology of an evolutionary programming approach with two new features: (1) dominant and recessive traits and (2) intron and exon regions. These features form the basis of a specialized approach for evolutionary programming which might be able to be applied to new problem areas where evolutionary programming usually performs poorly. This specialized approach is applied to the well known problem of a series expansion, so that the results are easily * compared to a known solution, and that the influence of these additional mechanisms on the population of solutions can be studied. Results from implementing the new mechanisms individually and together are presented, and compared with a baseline evolutionary programming implementation