Thesis (2)

This report presents the synthesis of research that has been done on Genetic Algorithms in the FRACTALES group of INRIA since its creation in 1993. The use of Genetic Algorithms for the resolution of optimization problems that arise when dealing with fractal analysis of signals was an important step towards the efficient resolution of basic inverse problems such as inverse problems for IFS. In this report we present, among other applications, how Genetic algorithms and genetic programming techniques have been implemented in order to find solutions in a tractable computation time.

Furthermore, a theoretical investigation has been carried out concerning the deception for genetic algorithms with the help of “fractal” tools. A link has thus been established between some irregularity measures on the function to be optimized and its deception. A qualitative analysis of the influence of some of the GA parameters has been made and a chromosomal encoding evalutation tool has been designed.