Combining cartesian genetic programming with an estimation of distribution algorithm

This paper describes initial testing of a novel idea to combine a CGP with an EDA. In recent work a new improved crossover technique was successfully applied to a CGP. To implement the new method meant changing the traditional CGP representation. The new representation developed in that work lends itself very nicely to some probability distribution being implemented. The work in this paper has investigated this idea of incoporating estimated probability distributions into the new CGP method with crossover. Categories and Subject Descriptors I.2.2 [Artificial Intelligence]: Automatic Programming— Program synthesis; I.2.8 [Artificial Intelligence]: Problem Solving, Control Methods and Search General Terms Algorithms, Design, Performance Keywords Cartesian Genetic Programming, optimization, crossover techniques

• Type: Conference
• Year: 2008
• Where: GECCO
• Authors: Janet Clegg
• Keywords: Crossover Techniques, GECCO 2008, Optimization, Probability Distributions, Traditional Cgp Representation