2014 | OriginalPaper | Buchkapitel
Particle Swarm Optimization Combined with Query-Based Learning Using MapReduce
verfasst von : Jeng-Wei Lin, Wen-Chun Chi, Ray-I Chang
Erschienen in: Future Information Technology
Verlag: Springer Berlin Heidelberg
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Particle swarm optimization (PSO) has shown its effectiveness to solve many complex optimization problems. However, PSO sometimes may fall into a local optimal solution rather than the global optimal solution of a given problem. For large-scale optimization problems, several parallelized PSOs have been proposed in the literature. In this paper, a query-based learning (QBL) approach is adopted to help PSO jump out of a local optimum. An Oracle is introduced that answers PSO whether there are too many particles in a flat region of the solution space. If yes, the algorithm will redistribute some of the particles in the flat region to somewhere else. A parallelized implementation, referred to as MRPSO-QBL, is developed in the Apache Hadoop MapReduce framework, as the framework provides a simpler and better parallel programming paradigm. The experiment results on several benchmark functions have demonstrated that MRPSO-QBL can find better solutions and converge faster.