Evolutionary optimisation of noisy multi-objective problems using confidence-based dynamic resampling.
European Journal of Operational
AbstractMany real-world optimisation problems approached by evolutionary algorithms are subject to noise. When noise is present, the evolutionary selection process may become unstable and the convergence of the optimisation adversely affected. In this paper, we present a new technique that efficiently deals with noise in multi-objective optimisation. This technique aims at preventing the propagation of inferior solutions in the evolutionary selection due to noisy objective values. This is done by using an iterative resampling procedure that reduces the noise until the likelihood of selecting the correct solution reaches a given confidence level. To achieve an efficient utilisation of resources, the number of samples used per solution varies based on the amount of noise in the present area of the search space. The proposed algorithm is evaluated on the ZDT benchmark problems and two complex real-world problems of manufacturing optimisation. The first realworld problem concerns the optimisation of engine component manufacturing in aviation industry, while the second real-world problem concerns the optimisation of a camshaft machining line in automotive industry. The results from the optimisations indicate that the proposed technique is successful in reducing noise, and it competes successfully with other noise handling techniques.
A parallel surrogate-assisted multi-objective evolutionary algorithm for computationally expensive optimization problems. Computation, CEC 2008 (pp. 3177-3184
In: 2008 IEEE Congress on Evolutionary
Preprint of paper published in Proceedings of IEEE Congress on Evolutionary Computation, 2008Abstract-This paper presents a new efficient multiobjective evolutionary algorithm for solving computationallyintensive optimization problems. To support a high degree of parallelism, the algorithm is based on a steady-state design. For improved efficiency the algorithm utilizes a surrogate to identify promising candidate solutions and filter out poor ones. To handle the uncertainties associated with the approximative surrogate evaluations, a new method for multi-objective optimization is described which is generally applicable to all surrogate techniques. In this method, basically, surrogate objective values assigned to offspring are adjusted to consider the error of the surrogate. The algorithm is evaluated on the ZDT benchmark functions and on a real-world problem of manufacturing optimization. In assessing the performance of the algorithm, a new performance metric is suggested that combines convergence and diversity into one single measure. Results from both the benchmark experiments and the realworld test case indicate the potential of the proposed algorithm.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.