SonOpt: understanding the behaviour of bi-objective population-based optimisation algorithms through sound

Abstract We present an extension of SonOpt, the first ever openly available tool for sonification bi-objective population-based optimisation algorithms. SonOpt has already introduced benefits on understanding algorithmic behaviour by proposing use sound as a medium process monitoring The edition utilised two different paths to provide information convergence, population diversity, recurrence objective values across consecutive generations and shape approximation set. provides further insight through introduction third path, which involves hypervolume contributions facilitate relative importance non-dominated solutions. Using generation approach than existing ones, this newly proposed path utilizes pitch deviations highlight distribution To demonstrate we compare sonic results obtained from popular multi-objective algorithms, Non-Dominated Sorting Genetic Algorithm (NSGA-II) Multi-Objective Evolutionary based Decomposition (MOEA/D), Multi-objective Random Search (MRS) baseline. three algorithms are applied numerous test problems showcase how can reveal various aspects that may not be obvious visualisation alone. is download at https://github.com/tasos-a/SonOpt-2.0 .