• Home
• Goals
• Researchers
• Publications
• Deliverables

GOALS

• Multi-objective optimization techniques in parallel and distributed heterogeneous architectures with a balanced use of computing, communication and storage, aiming to reduce execution times and energy consumption, and to improve the quality of the solutions found. Thus, new parallel evolutionary procedures based on sub-populations will be developed, as it has been identified in the TIN2012-32039 project as a plausible approach to high-dimensionality problems. Also, new ways of cooperation, with asynchronous communication between processors, data access locality, efficient data storage distribution taking advantage of the features of certain distributed filesystems, will be explored. Furthermore, pool based implementations of co-evolutive cooperation techniques for multi-objective optimization will be considered and compared in order to improve the fault tolerance of applications.
• Design of multi-objective approaches for data analysis problems such as feature selection, classification, and clustering on large and/or high dimensionality data sets. These multi-objective approaches will be implemented using the parallel code developed in the previous point.
• Development of socio-economic relevance applications on biomedical engineering, mainly related to neuroengineering, rehabilitation technologies and medical images processing, in which the research team has previous experience. The goal is to facilitate the development of applications involving continuous bio-signals monitoring, which imply the identification of concrete features and the reaction in real-time. This is where high performance computing will make possible the approach to the high dimensionality optimization and multi-objective problems presented.