Observatorio de I+D+i UPM

Memorias de investigación
Research Project:
PARALLEL EVOLUTIONARY OPTIMIZATION FOR HIGHLY COMPUTING .....
Year:2010
Research Areas
  • Artificial intelligence
Information
Abstract
This project is directed towards the development of new techniques for the heuristic optimization, applicable to problems computationally very demanding. Even though the techniques proposed by this project will be valid for a wide range of optimization problems, two real scenarios will be applied in order to prove their validity; in the areas of industrial design and data mining. The heuristic optimization techniques based in populations have already proved their usefulness in real problems within the industrial sector, and also as algorithm wrappers for machine learning. Nevertheless, with some of those applications the high number of candidate solutions and the computational cost of their evaluation make them too heavy. This type of problems appears in industrial design processes, based in simulation techniques for prototype evaluation. From a computational point of view, the simulation is very costly and the number of possible simulations is limited by the computational resources and the time invested on the process. Same thing happens with the application of specific combined techniques (evolutionary + machine learning). There, each solution means the execution of a data mining algorithm, (once or several times), parametrized by the proposed solution. In this sense, this project intends to explore the use of parallel optimization techniques, based in evolutionary algorithms. This can be useful for applications whose evaluation functions (fitness calculation) are computationally heavy, and where the current parallel techniques are not adequate. Our approach suggests exploring complementary strategies, in order to improve the results and, above all, to reduce the time needed for the optimization. With regard to the field of industrial design, the project will tackle the design optimisation of structures made of several layers of composite materials, in which the resultant weight is one of the key quality factors. For this, collaboration with the company Airbus-Spain will be undertaken. As for the data analysis techniques, our work will be focused in the machine learning field applied to biology. To that end, we will have the invaluable collaboration of distinguished groups of researchers [Alex A. Freitas from the University of Kent (UK), FazelFamili from the National Research Council (Canada) and Guillaume Beslon from the INSA-Lyon (France)]. Those researchers groups, whom we have close collaboration, are currently working on the combination of evolutionary techniques, statistics, data mining processes and machine learning algorithms.
International
No
Project type
Proyectos y convenios en convocatorias públicas competitivas
Company
MEC / Plan nacional de investigación
Entity Nationality
Sin nacionalidad
Entity size
Pequeña Empresa (11-50)
Granting date
Participants
  • Director: Jose Maria Peña Sanchez (UPM)
Research Group, Departaments and Institutes related
  • Creador: Grupo de Investigación: COMPUTATIONAL INTELLIGENCE GROUP
  • Centro o Instituto I+D+i: Centro de tecnología Biomédica CTB
S2i 2020 Observatorio de investigación @ UPM con la colaboración del Consejo Social UPM
Cofinanciación del MINECO en el marco del Programa INNCIDE 2011 (OTR-2011-0236)
Cofinanciación del MINECO en el marco del Programa INNPACTO (IPT-020000-2010-22)