Handbook of Metaheuristic Algorithms

Chapter Seventeen: Pattern reduction

Abstract

An efficient framework for reducing the computation time of metaheuristic algorithms, called pattern reduction (PR), is described in this chapter. Compared with parallel technologies that can only be used to reduce the “response time” of metaheuristic algorithms, PR can be used to reduce the “computation time” of metaheuristic algorithms by eliminating redundant computations during the convergence process. Motivated by the observation and assumption that subsolutions of metaheuristic algorithms will reach their final states at different times, the major task of PR is to detect and remove subsolutions that have reached their final states and for which any further computations are thus essentially redundant. ...