Archivers for Multi-objective Optimization

Multi-objective optimization problems have sets of optimal solutions rather than a single answer. An archiver is the component of an evolutionary algorithm that decides which solutions to keep during the search. This project studies the theoretical properties of archivers — convergence, approximation quality, limit behavior — and develops new archiving strategies with provable guarantees. The work is summarized in a Springer monograph and has produced several publications in IEEE Transactions on Evolutionary Computation.

In collaboration with Dr. Oliver Schütze, CINVESTAV-IPN.