Evoperf Project Website
EvoPerf is a University of Luxembourg project.
This project takes place in the ILIAS laboratory of the Computer
Science and Communications (CSC) Research unit. By choosing a
system biomedicine application in collaboration with the Luxembourg
Centre for Systems Biomedicine (LCSB), EvoPerf aims at
performing interdisciplinary research.
EvoPerf aims at providing the bases for robust and performance guaranteed evolutionary computations. Such methods have a large spectrum of applications.
Many of the real world problems are intractable (NP-Hard), whereas different approaches exist, including problem relaxation or local approaches. However most techniques rely on stochastics to explore different starting points (iterated gradient) or diversify the search (meta-heuristics). More than 15 years ago proof of convergences of stochastic based approaches were provided, e.g. for simulated annealing and genetic algorithms. But, most of the research on genetic type particle algorithms, evolutionary computation and/or Monte Carlo literature seems to be developed with no visible connections to the physical or the mathematical sides of this field.
In the current project we intend to extend the approach to cutting edge parallel and robust multi-objective particle algorithms (differential evolution, cellular genetic algorithms), both at a theoretical and implementation level. Validation will be carried on a cutting edge system biomedicine problem, providing new models/tools for protein folding. The obtained results will help in the development of new and more efficient drugs. In addition, EvoPerf's algorithms, platforms and middleware, will benefit to other fields where robust and large-scale evolutionary computation are required.
