Section: New Results
The main idea is the development of a methodology to generate a reasonable set of approximated Pareto' solutions (closed to the best achievable solutions). Especially, we have applied this methodology to better take into account users' criteria than the other existing methods offer. We have also studied the problem of selection of best algorithms in a portfolio. This research axis is currently enforced by the INRIA postdoc position of Joachim Lepping where we have started to include a learning process to select the best algorithm on a given instance.