2018/06/04 by Marcella Scoczynski Ribeiro Martins, Mohamed El Yafrani, Martins, Marcella S. R. +9
Computer Science · #Advanced Multi-Objective Optimization Algorithms #Artificial Intelligence (cs.AI) #Data Structures and Algorithms (cs.DS) #Evolutionary Algorithms and Applications #FOS: Computer and information sciences #Metaheuristic Optimization Algorithms Research
paper · pdf · doi:10.48550/arxiv.1806.09935
openalex publication_date 2018/06/04 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Fitness landscape analysis investigates features with a high influence on the\nperformance of optimization algorithms, aiming to take advantage of the\naddressed problem characteristics. In this work, a fitness landscape analysis\nusing problem features is performed for a Multi-objective Bayesian Optimization\nAlgorithm (mBOA) on instances of MNK-landscape problem for 2, 3, 5 and 8\nobjectives. We also compare the results of mBOA with those provided by NSGA-III\nthrough the analysis of their estimated runtime necessary to identify an\napproximation of the Pareto front. Moreover, in order to scrutinize the\nprobabilistic graphic model obtained by mBOA, the Pareto front is examined\naccording to a probabilistic view. The fitness landscape study shows that mBOA\nis moderately or loosely influenced by some problem features, according to a\nsimple and a multiple linear regression model, which is being proposed to\npredict the algorithms performance in terms of the estimated runtime. Besides,\nwe conclude that the analysis of the probabilistic graphic model produced at\nthe end of evolution can be useful to understand the convergence and diversity\nperformances of the proposed approach.\n