Identification of metrics suitable for determining the features of real-world optimisation problems. (February 2022)

• Record Type: Journal Article
• Authors: Zhu, S.; Maier, H.R.; Zecchin, A.C.
• Is Part Of: Environmental modelling & software. Volume 148(2022)
• Publication Date: 2022-02
• Subjects: ELA Exploratory Landscape Analysis -- BBOB Black-Box Optimisation Benchmark -- ANN Artificial Neural Network -- ICoFS Information Content of Fitness Sequences; Optimisation -- Calibration -- Fitness landscape -- Error function -- Exploratory landscape analysis (ELA) -- Evolutionary algorithms; Environmental monitoring -- Computer programs -- Periodicals; Ecology -- Computer simulation -- Periodicals; Digital computer simulation -- Periodicals; Computer software -- Periodicals; Environmental Monitoring -- Periodicals; Computer Simulation -- Periodicals; Environnement -- Surveillance -- Logiciels -- Périodiques; Écologie -- Simulation, Méthodes de -- Périodiques; Simulation par ordinateur -- Périodiques; Logiciels -- Périodiques; Computer software; Digital computer simulation; Ecology -- Computer simulation; Environmental monitoring -- Computer programs; Periodicals; Electronic journals; 363.70015118
• DOI: 10.1016/j.envsoft.2021.105281
• Languages: English
• ISSNs: 1364-8152

Abstract

Abstract : Optimisation methods are applied increasingly to environmental problems. Much research in this area is concerned with the behaviour of optimisation algorithms, however, the effectiveness of these algorithms is also a function of the features of the problem being solved. Although a number of metrics have been developed to quantify these features, they have not been applied to environmental problems. The primary reason for this is that the computational cost associated with the calculation of many of these metrics increases significantly with problem size, making them unsuitable for real-world problems. In this paper, 28 fitness landscape metrics that have low dependence on problem size are identified through extensive computational experiments on a range of benchmark functions and testing on a number of environmental modelling problems. These metrics can be applied to real-world optimisation problems in a computationally efficient manner to better understand their features and determine which optimisation algorithms are most suitable. Highlights: Optimisation are used increasingly for a range of environmental problems. Fitness landscape metrics help to understand the features of optimisation problems. Many fitness landscape metrics are too time consuming for real-world problems. 28 fitness landscape metrics that can be applied to real-world problems are found. These metrics provide insight into the selection of optimisation algorithms.