Genetic Programming Based on an Adaptive Regularization Method

A proper model of a bioprocess is very important for the development of industrial bioprocesses. Here, genetic programming (GP) and genetic algorithm (GA) are used to model the Avermectin fermentation process. To get more accuracy model without losing generalization, a regularization term is integrated into the fitness function and an adaptive method to optimize regularization parameter is proposed to balance training accuracy and the curvature of a nonlinear model. Furthermore, a new protected approach is proposed and experiments show that with the method, the amount the undesired sharp changes in the predicting value decreases largely