A parallel framework for multi-objective evolutionary optimization

This work focuses on the development of a parallel framework method to improve the effectiveness and the efficiency of the obtained solutions by Multi-objective Evolutionary Algorithms. Specifically, a parallel architecture based on JavaSpaces technology and an island paradigm model is proposed and tested on two important and complex computational problems: The Protein Structure Prediction and the Task based Navy´s Sailor Assignment problems. An experimental framework is developed in order to test the proposed parallel framework. Particularly, the framework is tested using real-world data belonging to the TSAP and PSP problem. Furthermore, new insights are obtained about modeling these problems as parallel Multi-objective Evolutionary Algorithms.