A comprehensive survey of fitness approximation in evolutionary computation

By Yaochu Jin. Soft Computing - A Fusion of Foundations, Methodologies and Applications, 2005

1,365 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/e1bbd52a6de0b0d7b4f513a88c617d47f52d2095

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Surrogate-Assisted Cooperative Swarm Optimization of High-Dimensional Expensive Problems (2017, methodology): “Table V presents the computation time of the algorithms under comparison averaged over 20 independent runs when a maximum of 1000 FEs using the real fitness function is allowed.”
A Gaussian Process Surrogate Model Assisted Evolutionary Algorithm for Medium Scale Expensive Optimization Problems (2014, methodology): “Among them, GP modeling, response surface methods, artificial neural networks, support vector machines, and radial basis functions exhibit good performances and are widely used [7], [8], [10], [12]–[16].”
Adaptive sequentially space‐filling metamodeling applied in optimal water quantity allocation at basin scale (2010, methodology): “Computational frameworks for integrating single‐point approximation models and EAs were proposed by Nair and Keane [2001]. Farhang‐Mehr and Azarm [2005] used a sequential approach with adaptation to irregularities in the response behavior for Bayesian metamodeling in engineering design simulations.”
Generalizing Surrogate-Assisted Evolutionary Computation (2010, methodology): “Among the approximation models, polynomial regression (PR), also known as response surface methodology (RSM), support vector machine (SVM), artificial neural networks (ANNs), radial basis function (RBF), and Gaussian process (GP), also referred to as Kriging or design and analysis of computer experi”
A survey on metaheuristics for stochastic combinatorial optimization (2009, methodology): “…Guo 2004) EC algorithms use the exactly computable objective function as it is, even if computationally expensive, while in the second group (Bianchi et al. 2004, 2006), and references cited in Jin (2005) EC exploits also computationally more efficient objective function (fitness) approximations.”
Hybrid Genetic Algorithms: A Review (2006, background): “Jin [59] provides a comprehensive survey on fitness approximation techniques.”
Evolutionary Bilevel Optimization for Complex Control Tasks (2015, methodology): “Numerous models have been proposed, including both parametric and nonparametric ones [27, 28, 29, 30].”
Local Approximation of Difference Evaluation Functions (2016, neutral): “This fitness approximation is typically used because the number of fitness function evaluations dominates the optimization cost in evolutionary algorithms, and fitness approximation typically decreases time to convergence [13, 17].”
Constraint-handling in nature-inspired numerical optimization: Past, present and future (2011, background): “In spite of the fact that fitness approximation methods have been extensively applied to unconstrained optimization problems [211,212], their use in CNOPs remains scarce.”
Infill sampling criteria for surrogate-based optimization with constraint handling (2012, methodology): “Specific reviews of approximation methods in evolutionary algorithms can be found in Jin (2005) and Lim et al.”
A Random Forest-Assisted Evolutionary Algorithm for Data-Driven Constrained Multiobjective Combinatorial Optimization of Trauma Systems (2020, methodology): “The use of EAs to solve combinatorial optimization becomes less practical, however, when fitness evaluations of candidate solutions rely on time-consuming numerical simulations or expensive physical experiments rather than computationally cheap analytic functions [4].”
A Tutorial On the design, experimentation and application of metaheuristic algorithms to real-World optimization problems (2021, methodology): “…paradigm resort to data-based regression techniques to infer the relationship between the decision variables and their objective function value, so that when learned, the evaluation of new candidates for the problem at hand can be eﬃciently performed by querying the trained regression model [145].”
On Using Surrogates with Genetic Programming (2015, methodology): “The basic idea here is to minimize the number of expensive fitness evaluations by replacing them with a computationally cheaper function as far as possible.”
Parameter tuning for configuring and analyzing evolutionary algorithms (2011, methodology): “Using meta models or surrogate models is a well established approach for the optimization of computationally expensive problems by (evolutionary) search [41,42].”
Evolutionary Computation Meets Machine Learning: A Survey (2011, methodology): “Among various modeling methods, the ANN technique (for single objective function [18] and for multi-objective function [19]), statistical methods [20], polynomial regression [21], radial basis function interpolation [22], and Markov fitness model [23], to name just a few, have been incorporated into”
Is it worth to approximate fitness by machine learning?: investigation on the extensibility according to problem size (2018, background): “In these cases, It is vital to build a model that approximates fitness and perform research on possibility of approximation [1–3].”
A Classification-Based Surrogate-Assisted Evolutionary Algorithm for Expensive Many-Objective Optimization (2019, background): “locating the global optimum or global Pareto front of highdimensional multimodal problems becomes hardly tractable as an accurate approximation of the fitness landscape will usually require a very large number of FEs [25].”