August 21, 2013


August 9-11, 2013, Reykjavik University, Iceland — on the at Second International Workshop on Advances in Simulation-Driven Optimization  and Modeling (ASDOM 2013) DATADVANCE presented new multi-objective surrogate based optimization approach. The results hown promising performance of new approach for optimization problems with calculational expensive objectives and constraints.

Alexis Popselov in hit talk "Multi-objective programming: adaptive surrogate based approach with Chebyshev scalarization" presented a promising optimization technique aimed to solved multi-objective optimization problems with computationally expensive models where strict limitations on amount of evaluations of models are imposed.

The key components of the proposed algorithm are adaptive scalarization, which allows achieving uniformity of approximation of Pareto frontier, and and surrogate models which tie together independent scalarized subtasks. 

Obtained results show that new technique allows efficiently solving black-box problems using relatively small amount of evaluations.