Multi-objective parameter-less population pyramid in solving the real-world and theoretical problems

Many real-world problems are notoriously multi-objective and NP-hard. Hence, there is a constant striving for optimizers capable of solving such problems effectively. In this paper, we examine the Multi-Objective Parameter-less Population Pyramid (MO-P3). MO-P3 is based on the Parameter-less Population Pyramid (P3) that was dedicated to solving single-objective problems. P3 employs linkage learning to decompose the problem and uses this information during its run. P3 maintains many different linkage information sets, which is the key to effectively solve the problems of the overlapping nature, i.e., the problems whose variables form a large and complicated network of dependencies rather than additively separable blocks. MO-P3 inherits the features of its predecessor and employs both linkage learning and linkage diversity maintenance to effectively solve hard multi-objective problems, which includes both: well-known test problems and NP-hard real-world problems.