This work proposes a new technique for the treatment of Multi-objective Optimization Problems (MOPs), based on the extension of a socio-cultural algorithm, the Method of Musical Composition (MMC). The MMC uses a society of agents, called composers, who have their own creative ability, maintain a memory of their previous artwork and are also able to exchange information.
According to this analogy, a decomposition approach implemented through a Tchebycheff function is adapted, assigning each composer to the solution of a particular scalar sub-problem. Agents with similar parameterization of the original MOP may share their solutions. Furthermore, the generation of new tunes was modified, using the Differential Evolution mutation operator. Computational experiments performed on the ZDT and DTLZ test suite highlight the promising performances obtained by the resulting MO-MMC algorithm, when compared with the NSGA-II, MOEA/D and two swarm intelligence based techniques.