Training Complex Decision Support Systems with Differential Evolution Enhanced by Locally Linear Embedding

摘要

This paper aims at improving the training process of complex decision support systems, where evolutionary algorithms are used to integrate a large number of decision rules in a form of a weighted average. It proposes an enhancement of Differential Evolution by Locally Linear Embedding to process objective functions with correlated variables, which focuses on detecting local dependencies among variables of the objective function by analyzing the manifold in the search space that contains the current population and transforming it to a reduced search space. Experiments performed on some popular benchmark functions as well as on a financial decision support system confirm that the method may significantly improve the search process in the case of objective functions with a large number of variables, which usually occur in many practical applications.