Constructing dense fuzzy systems by adaptive scheduling of optimization algorithms

In this paper dense fuzzy rule based systems are constructed for solving machine learning problems. During the knowledge extraction process a scheduling approach is applied, which adaptively switches between the different optimization algorithms based on their convergence speed in the phases of the learning process, i.e. according to their respective local efficiency. The scheduled optimization techniques are evolutionary algorithms that have shown efficiency in the construction of dense fuzzy rule based systems in previous investigations. Simulations runs are executed on standard benchmark data sets in order to evaluate the established fuzzy rule based learning system and to compare it to fuzzy systems built up using the same optimization methods without the scheduling approach.