Improving the Evolutionary Coding for Machine Learning Tasks

Abstract

The most influential factors in the quality of the solutions found by an evolutionary algorithm are a correct coding of the search space and an appropriate evaluation function of the potential solutions. The coding of the search space for the obtaining of decision rules is approached, i.e., the representation of the individuals of the genetic population. Two new methods for encoding discrete and continuous attributes are presented. Our “natural coding” uses one gene per attribute (continuous or discrete) leading to a reduction in the search space. Genetic operators for this approached natural coding are formally described and the reduction of the size of the search space is analysed for several databases from the UCI machine learning repository.